Frameworks Are Not Neutral

When Viacom acquired Pluto TV, the director of product and I were ready to take a media asset management system (MAM) to RFP.

We had spent months getting there. Pluto had scaled remarkably far without a MAM, but the limits were becoming clear. We had written the business case, developed functional and technical requirements, built a proposed metadata model, and secured review and approval from stakeholders across the company. We had surveyed the market and identified three likely vendors. The work was not finished, but the problem was understood and the next step was clear.

Then the acquisition closed.

Viacom wanted assurance that the approach would work. That was reasonable: a parent company was evaluating a significant technology decision inside a business it had just acquired. But instead of reviewing the work we had done, Viacom asked us to pause the project for a 12-week discovery engagement with an outside consulting team.

The practical requirement was that we fit Pluto’s operation and our proposed solution into the consultants’ framework so the consultants could explain it back to Viacom.

For roughly six weeks, two directors spent most of each day teaching the consultants how Pluto’s media supply chain worked. The consultants took notes. At the end of most days, we reviewed those notes, found misunderstandings, and corrected them the next morning. The MAM project stalled while the people who knew the operation best translated it for people who had been hired to translate it for our new owners.

Half way through, I went to the executive sponsor at Viacom and asked him to lift the requirement. He agreed. We were able to move forward, but only after losing six weeks. I later learned that the consulting work itself had cost about $40,000. That figure did not include the time of two directors, the delayed RFP, or the opportunity cost of stopping a company that Viacom had acquired in part because it could move quickly.

We had hoped for a rigorous review of our plan. What we received was a process designed to train the consultants.

Frameworks can make complex operations easier to discuss, and outside expertise can provide challenge, perspective, and skills a company does not have. But neither is neutral. A framework determines what becomes visible, what gets classified as noise, and what kind of answer appears reasonable and congruent to corporate strategy. The firm you hire to help with your media supply chain will bring a framework, and that framework will alter how leadership understands the problem.

That can be useful. It can also be expensive.

Subscribe now

A framework is a form of compression

A framework helps people decide what to look at. It groups related activities, names common capabilities, and proposes relationships among them. In a streaming value chain, it might connect content creation, acquisition, rights, asset management, processing, programming, distribution, monetization, and audience behavior.

This compression is why frameworks make communication easier. A clean diagram gives executives a shared vocabulary and a picture they can discuss. It puts boundaries around an operation that may cross dozens of teams, systems, vendors, and decisions.

But making complexity legible does not remove the complexity. It edits some of it out.

The clean boxes cannot fully represent conflicting rights records, incomplete deliveries, undocumented vendor knowledge, inherited technology, regional differences, one-off commercial commitments, or the person who knows which workaround will keep a launch on schedule. Yet those details often carry the operating risk. A framework can make the operation appear more orderly at exactly the moment leaders most need to understand its exceptions.

This is why the word *framework* needs qualification. An industry framework describes common roles or capabilities. A consulting framework structures an engagement. An improvement framework provides a method for investigating performance. An operating model defines how a particular company converts strategy into recurring decisions and work.

These tools overlap, but they are not interchangeable.

An operating model must eventually address ownership, governance, processes, systems, talent, incentives, measures, and decision rights. A value-chain diagram may help a company begin that work. It cannot complete it.

APQC’s Process Classification Framework for broadcasting makes the distinction concrete. It provides a common language for organizing processes and supporting comparison and benchmarking. That can be a useful starting point, particularly when different parts of a company use different words for similar work. But a taxonomy of processes is not an account of how one company’s rights, assets, systems, decisions, and exceptions interact. It can tell leaders which activities may exist without telling them how the operation should be designed.

Every framework carries a point of view

A cloud-oriented framework will make cloud opportunities easier to see. A commercial lifecycle will foreground monetization. A technology framework will expose architecture and integration gaps. A process-improvement method will look for variation, delay, and defects.

The framework influences the questions long before anyone reaches the answers. Choosing the lens is already a strategic decision.

Accenture and the World Economic Forum developed a Media, Sports and Platforms Value Map to represent major roles in content creation and distribution and the movement of value among them. Accenture describes it as providing a directional understanding, with the effects varying by company and circumstance. A directional map can help leaders orient themselves. It does not prescribe how an individual streaming company should operate.

A second Accenture model divides the media supply chain into contribution and distribution. Its cloud orientation makes architectural dependencies visible, particularly for a company considering cloud transformation. That orientation also determines what receives attention.

Deloitte examines the content lifecycle through five opportunities: greenlighting strategy, distribution strategy, production efficiency, audience engagement, and advertising effectiveness. This commercial-value lens helps leaders consider how choices across the lifecycle affect content return. It is less suited to documenting the operational mechanics of getting a particular title from contract through monetization.

Qvest publishes an implementation approach organized around Strategy, Core, Accelerate, and Align. It moves from current- and future-state analysis through minimum viable product development and iterative expansion. This is closer to a transformation method than a universal model of a media value chain.

Each model can be useful. Each has also been designed to make some things easier to see than others. If a company hires a consulting partner without understanding that point of view, it may adopt the partner’s definition of the problem before deciding whether it is the right one.

Six Sigma needs a process that is ready to improve

Six Sigma enters these conversations because it offers discipline. Its DMAIC method defines the problem, measures the current process, analyzes causes, improves performance, and establishes controls. The American Society for Quality describes DMAIC as a method for improving an existing process that does not meet performance standards or customer expectations. It is especially suited to incremental improvement when performance can be measured and the causes of variation can be investigated.

That makes Six Sigma valuable under the right conditions. The process must repeat often enough to measure. The desired outcome must be known. The organization needs a meaningful definition of a defect and enough trustworthy data to establish a baseline.

Many streaming operations are not there yet.

Variation can represent a defect. It can also represent a different rights model, content type, territory, partner obligation, or revenue strategy. A team can reduce processing time by pushing unresolved problems downstream. It can improve an on-time delivery metric by excluding the packages requiring remediation. It can standardize a workflow whose output no longer serves the business.

Reducing variation in the wrong process produces a consistent wrong result.

Before applying Six Sigma, leaders need to determine whether they have a performance problem inside a valid process or an operating-model problem that requires a different design. Measurement can improve the first. It may merely harden the second.

The choice of method should follow a diagnosis of the problem, not precede it. The Cynefin framework distinguishes among clear, complicated, complex, chaotic, and confused environments, each of which calls for a different form of action. A repeatable transcoding workflow may be suitable for measurement and standardization. The design of accountability across rights, programming, product, and operations may be a complex problem whose solution has to emerge through testing. Applying the same improvement method to both mistakes consistency for rigor.

Streaming value chains share principles, not operating models

Streaming companies encounter many of the same underlying requirements. Rights must become actionable instructions. Assets and metadata must arrive and remain connected. Content has to be validated, transformed, packaged, programmed, and delivered. Exceptions need resolution. Quality, cost, timing, and performance need to be measured.

Even the phrase *supply chain* introduces a point of view. The Supply Chain Operations Reference model, or SCOR, organizes supply-chain activity around Orchestrate, Plan, Order, Source, Transform, Fulfill, and Return. That structure is flexible and useful for many industries, but it reflects assumptions developed around the fulfillment of customer orders. Media assets are not depleted when they are delivered. They can support many products at once, and their usability depends on rights, metadata, programming, and commercial context. Applying a conventional supply-chain lens to media can therefore make certain forms of flow and efficiency highly visible while making exploitation, optionality, and editorial judgment harder to see.

The similarities become less reliable as the view moves closer to the work.

A subscription service built around premium originals has a different operating burden from a FAST platform ingesting large volumes of acquired library content. A global service managing dozens of language versions has different controls from a domestic service. A company licensing titles one by one has different rights and cost structures from one acquiring channels or entire catalogs. Live sports, news, films, episodic television, and short-form content create different failure modes.

The organization changes the chain too. Rights may sit with legal, acquisitions, operations, or a dedicated rights team. Metadata ownership may be clear, divided, or absent. Distribution requirements may live in a product system, a spreadsheet, a vendor portal, or someone’s memory. The same conceptual capability can be performed through entirely different combinations of people and technology.

A generic model can show that rights management belongs in the system. It cannot reveal that a contractual restriction is stored in one database while the person building the schedule works from another.

That knowledge lives inside the operation. It is also the knowledge most vulnerable to being compressed away when an outside framework becomes the primary means of explaining the business.

Match the engagement to the maturity of the work

The Pluto engagement went wrong because the method did not match the maturity of the project. Maturity in this context describes how far a company has progressed in defining the problem and its proposed response. It does not mean the underlying operation is efficient.

Discovery would have made sense if the problem were poorly understood, the requirements had not been gathered, or stakeholders had never reconciled their needs. By the time the consultants arrived, those activities were substantially complete. The open question was whether the proposed design was sound.

A focused validation could have tested whether the metadata model supported the stated use cases, whether the requirements were complete, whether the architecture could scale, and whether the vendor shortlist missed a material capability or risk. A red-team review could have challenged our assumptions and identified the evidence that would require a change in direction.

Instead, the engagement restarted discovery because discovery was the consultants’ way of entering the problem. Our work had to be recast into their framework before it could become legible to the parent company. That was not simply an inefficient sequence of meetings. It shifted authority away from the people who understood the operation and toward the people who controlled its representation.

Outside validation can help when internal consensus is fragile, incentives are misaligned, or stakeholders cannot challenge one another safely. Consultants can also bring specialized knowledge, comparative experience, independent challenge, temporary capacity, and a structured way to resolve disagreement. But the mandate should be explicit, and the method should begin at the company’s actual starting point:

• An undefined strategy may require executive alignment and facilitated decision-making.

• A poorly understood operation may require discovery and current-state mapping.

• A mature internal design may require validation or a red-team review.

• A defined vendor direction may require selection diligence and implementation planning.

• A stable but underperforming process may benefit from a formal improvement method.

• A missing internal capability may justify targeted expertise or embedded execution support.

Buying discovery because it is the consulting firm’s standard first phase can be an expensive way to ignore how much the company already knows.

Count the cost of translation

Consulting fees are only the most visible part of an engagement’s cost.

The full cost includes the client labor required to service it, the decisions delayed while the work is underway, the operators diverted from execution, and the opportunity cost of teaching an outside team what the inside team has already learned. It also includes the risk that nuance will be lost as operating knowledge passes from practitioners to consultants and then from consultants to executives.

That last cost is easy to miss because translation is often the reason the consultants were hired. Leaders may trust an outside firm’s vocabulary, presentation style, or brand more than the messier explanations coming from inside the operation. The consultant becomes a bridge between the people doing the work and the people authorizing it.

Sometimes that bridge is necessary. But companies should recognize what they are buying. If the consultant’s framework becomes the only form in which leadership can understand the supply chain, the consultant is no longer merely analyzing the problem. The consultant is shaping the organization’s perception of it.

One safeguard is to select a consultant who can frame the problem through the company’s corporate strategy rather than forcing the strategy into a prefabricated model. That means more than aligning the engagement to a vision statement. A vision describes an intended future. A complete strategy explains where the company will compete, how it expects to win, which capabilities and investments matter most, what economic logic supports those choices, and which tradeoffs it is prepared to make. It also defines constraints, priorities, and decision principles that should shape the operating model.

The distinction matters. Two streaming companies may share a vision of reaching more viewers with better content and still require very different supply chains. One may compete through global scale and standardization. Another may win through local curation, rapid experimentation, or the ability to monetize fragmented libraries that other companies cannot operate efficiently. A framework that treats their desired end state as essentially the same will erase the strategic differences the operation exists to serve.

A strong outside partner should be able to use the corporate strategy as the governing frame, then choose or adapt other frameworks beneath it. The technology architecture, process taxonomy, improvement method, and measures should all be evaluated according to whether they enable the company’s actual choices. If the strategy is incomplete or contradictory, the consultant should surface that problem rather than quietly substituting its own assumptions.

Before hiring a firm, leaders should ask not only whether it has relevant experience, but how it will frame the work:

• How will it translate the company’s complete corporate strategy? Not only its vision, but also into operating-model choices?

• Which strategic priorities, constraints, and tradeoffs will govern its recommendations?

• What does its framework treat as the primary unit of analysis: technology, process, content, revenue, customer value, or organizational accountability?

• What assumptions does it make about standardization, centralization, and scale?

• How does it preserve exceptions and institutional knowledge rather than classify them as noise?

• How will it distinguish work that needs to be discovered from work that needs to be tested?

• What internal time will the engagement consume, and what will stop while that time is being spent?

• At the end, will the company understand its operation better, or merely understand the consultant’s description of it?

The firm should be selected partly for the framework it will bring and its willingness to adapt that framework to the strategy, not only for its reputation or staffing model.

Consultants cannot substitute for strategy

For the sake of accountability, I need to add that frameworks become especially attractive when leadership has not completed the harder work of setting direction.

A company should know what it is trying to accomplish, where it will compete, how it intends to win, what prevents it from getting there, and which principles will govern the choices ahead. It needs to align money, technology, people, and executive attention with those choices. It needs enough internal talent to execute, evaluate, and revise the plan. It also needs an accountable owner who will remain after the engagement ends.

Consultants cannot take ownership of the tradeoffs. Leadership still has to decide which consequences the company is willing to carry.

When that responsibility is outsourced, a framework can create the appearance of strategic progress. The boxes have names. The workstreams have colors. The roadmap has quarters. The consultants will be gone when the consequences come, but they are willing to come back when that happens – for a fee.

Let the framework disappear into the work

A good framework should become less visible as the operation matures.

Its useful ideas are translated into metadata rules, system behavior, workflows, controls, measures, decision rights, and clear ownership. Operators test its assumptions every time content arrives incomplete, a rights rule conflicts with a release plan, or a distribution partner changes its requirements. Its categories change when the business changes. Eventually, people may stop referring to the original diagram because the knowledge has been absorbed into the way the company runs.

This is different from discarding structure. Mature operations still need standards and controls. They also need a mechanism for revising them. The Lean concept of standardized work treats the documented process as a baseline for continuous improvement rather than a permanent answer.

The test is straightforward. Does the framework help the company make a better decision? Does it expose dependencies and exceptions? Can operators recognize their work in it? Does it clarify ownership, cost, and risk? Can the company revise it without commissioning another engagement?

Frameworks are useful because they leave things out. That is also why they are dangerous. Use them to make the operation discussable, but do not confuse the discussion tool with the operation itself. And choose outside partners with care: the framework they bring will influence not just how they organize the work, but how your organization comes to understand its own problem.

A framework has done its job when the company understands the operation well enough to stop organizing the work around the framework.

Research afterthought: other frameworks worth exploring

These frameworks, standards, and methods did not all fit the argument above, but they are useful reference points for anyone exploring media supply chains or technology-product operating models.

Media and supply chain

• EBU Digital Media Value Chain: An end-to-end model of the business objects and processes involved in commissioning, production, publication, distribution, consumption, audience response, and demand.

• MovieLabs 2030 Vision: Ten principles for the future of media creation, emphasizing cloud infrastructure, zero-trust security, and software-defined workflows.

• MovieLabs Ontology for Media Creation: A connected ontology, vocabulary, and visual language for representing the people, assets, tasks, and infrastructure involved in media creation.

• SMPTE Interoperable Master Format: A family of standards for storing and exchanging a high-quality master and its multiple versions across territories and platforms. It is narrower than an operating framework, but consequential for how a global content supply chain is designed.

• EBUCore and the EBU Class Conceptual Data Model: Reference data models for media information, asset management, production, distribution, rights, commercial, and regulatory domains.

• DPP: Industry specifications, research, and collaborative work spanning production, post-production, technology, and distribution. DPP is not a single operating framework, but its work has materially shaped common language and practices across the media supply chain.

• Lean Value-Stream Mapping: A method for visualizing the flow of work and information in order to identify delay, waste, and opportunities for improvement.

Technology and product

• Design Council Double Diamond: A product and service design model that alternates between divergent and convergent work: discovering the problem, defining it, developing possible responses, and delivering a solution.

• Scrum: A lightweight framework for developing and sustaining complex products through small, cross-functional teams and iterative delivery.

• ITIL: A service-management framework organized around the creation, delivery, support, and continual improvement of technology-enabled services.

• TOGAF: An enterprise-architecture framework and development method for connecting business needs with data, application, and technology architecture.

• Team Topologies: An organizational-design approach that treats team structure and interaction patterns as part of the architecture of technology delivery.

• Scaled Agile Framework: A system for coordinating product development, portfolio priorities, architecture, and delivery across many teams. Like the other frameworks listed here, its usefulness depends heavily on the problem, scale, and organizational context.

Some industries treat their supply chains as strategy. Apple ships new products every September at planetary scale, on the same day, in dozens of markets, with millions of units on shelves. That reliability is built on decades of investment in operational rigor across sourcing, manufacturing, distribution, and inventory management. Zara, the flagship brand of Inditex, can move a design from sketch to store in two to four weeks through a vertically integrated supply chain and in-house manufacturing in Spain and nearby countries. Costco carries roughly 3,700 SKUs against a typical grocery store’s 30,000 and turns inventory twelve times a year against an industry average of five to six, which is what makes the membership model work.

None of that is because these companies treated their supply chain operations as an afterthought or a technical detail. Those companies decided, at specific moments in their histories, that operations would be a strategic asset rather than a cost center. They built it that way and they defended it that way.

Streaming did not, and that difference is starting to matter more than it used to. The bill is finally coming due.

What the deficit looks like from the inside

Streaming, as an industry, does not lack operational talent. There are excellent operators inside almost every streaming platform in the market. There are heads of content operations, distribution leaders, metadata specialists, and platform engineers who could hold their own against operators in any industry. What streaming lacks is a shared operational discipline. Rigor exists in pockets. It rarely exists as an organizing principle.

The pattern I’ve watched, across close to two decades inside content operations for streaming and broadcast, is that most companies care about their operations. What varies is how evenly operational understanding and accountability are distributed inside the company. A given streamer might have a strong metadata governance program on paper and a weak tech stack behind it, or a mature FAST channel launch process that lives on the shoulders of a single vendor. When a business with fragile or immature operations pivots or iterates, the pipeline degrades. And because the operational layer was never fully realized, the executives who inherit these companies through acquisition, promotion, or restructuring often can’t see what they’re actually inheriting.

The operating layer at any media company is complex, often load-bearing in ways nobody has documented. The executives being asked to run streaming companies against increasingly punishing profit targets are working from an incomplete map. Some of them know the map is incomplete. Some don’t. Both cases produce measurable financial consequences.

The numbers

PureFacts, a revenue intelligence firm, has found that companies typically lose between one and five percent of annual revenue to operational leakage. For a mid-size streamer at $500 million in revenue, that is $5 to $25 million a year leaking from the supply chain without ever landing in the profit line. It is a small enough percentage to hide inside quarterly variance, and a large enough dollar figure to matter to investors.

The pattern gets more expensive at scale. Paramount Skydance’s pending acquisition of Warner Bros. Discovery would close the merged company with roughly $79 billion in net debt, in what has been described as the largest leveraged buyout in history. Paramount has committed to over $6 billion in synergies, driven by technology integration, streaming stack consolidation, and other operational streamlining. The Hollywood Reporter has noted the well-documented pattern that in leveraged buyouts of this size, acquirers typically capture less than half of the projected synergies. The interest expense alone on the combined debt could reach $5 to $6 billion a year, roughly half the entity’s projected EBITDA.

Paramount already missed Q4 2025 by ten cents per share, a 500 percent miss against consensus, driven by integration costs and operational headwinds that management appears to have underestimated. That is one specific quarter at one specific company. It also represents a pattern showing up across large media M&A. The synergy models assume that two operating layers can be merged into one at a specified pace and cost. The historical data on those assumptions is not encouraging.

What Wall Street is actually asking for

Wall Street has stopped rewarding streaming companies for subscriber growth in isolation. The evidence is direct and it comes from analysts who cover the sector every day.

Doug Creutz, senior research analyst at Cowen, told CNBC in April 2026: “At this point, no one is reporting subscriber numbers anymore; the focus has shifted entirely to profitability. And that is the benchmark against which these enterprises are evaluated. The question now is, can you achieve a 10% operating profit? A 15%?” Robert Fishman at MoffettNathanson has been making a related case for years and now argues that streaming is a viable business only for platforms that achieve sufficient scale.

The customer economics tighten the vice further. Netflix spends approximately $89 to acquire each new subscriber, and some streaming providers spend up to $200 per year in marketing per new customer. Retention costs roughly five to seven times less than acquisition. Every dollar of revenue leaking out through operational drag has to be re-earned through customer acquisition at those rates. Every FAST channel that takes three months to launch instead of two weeks is inventory that never generates revenue during the delay. Every metadata error that reaches the consumer is churn risk on a market that has already begun to lose subscribers to cost fatigue.

None of that is new. What is new is that the capital patience for the pattern has run out. The financial market previously willing to fund subscriber growth ahead of profitability is now asking a different question, and the answer requires operational rigor the industry never fully built.

Why the industry got here

Streaming grew up inside the Silicon Valley cultural moment that treated operations as friction. Move fast and break things. Ship it and figure the rest out later. Operational discipline, in that story, was the voice trying to slow the founder down. It was for people who wanted checklists, who couldn’t tolerate ambiguity, who did not understand that speed was the whole game.

That story produced some genuine casualties. WeWork, Theranos, and FTX each burned through significant capital with governance and operational failures that were visible internally long before they became public. The pattern also shaped how streaming built itself. The industry inherited the operational instincts of its early technology backers rather than the operational instincts of the broadcast industry that preceded it. Broadcast had its own operational deficits. Metadata was often treated as a compliance deliverable in a binder rather than an asset. But broadcast at least maintained certain kinds of workflow discipline, driven by the constraints of a linear schedule and standards bodies with enforceable authority.

Streaming replaced those constraints with technology-first thinking. That produced enormous creative freedom and enabled the growth curve of the last fifteen years. It also produced organizations where CEOs can’t explain their own content supply chain, where vendor relationships often substitute for internal strategy and core competency, and where the connective tissue between strategy and execution frequently gets built by whoever is closest to the immediate problem rather than by design.

The companies that survived the Silicon Valley boom and compounded, the Apples, the Costcos, the Amazons, and Netflix at its operational best, all made deliberate investments in operational rigor at specific points in their histories. They did it because someone at the top of those companies decided supply chain operations was part of their strategy. Streaming, as a broader industry, has not consistently made that decision.

What actually accelerates growth

Operational rigor, done well, accelerates growth. That is the counterintuitive point most executives in the industry are being asked to internalize now, often under conditions where the stakes make it hard to hear.

The streaming companies that will compound over the next decade are the ones that treat the operational layer the way Apple treats sourcing or Inditex treats production. That doesn’t require reinventing the pipeline. It requires making it legible. Executives at those companies will need to know their own pipelines end to end. They will need to know which handoffs are needlessly manual, which decisions are undocumented, which vendors are shaping the operating model without accountability, and which teams are compensating for systems that nobody has fully fixed. They will need to know where the money leaks and to have a plan for closing the leak.

Most streaming leadership doesn’t have that map yet. Some of them are aware of the gap. Some are surrounded by people who know pieces of it but haven’t been asked to assemble them. Some are inheriting the gap from predecessors and are still learning where the load-bearing walls are. In each of those cases, the situation is solvable, but the solution requires attention rather than another platform migration.

Operational rigor at the senior leadership level is a solvable problem. It doesn’t require a fifty-million-dollar technology overhaul. It requires investment of executive time in walking the pipeline, honest diagnosis, and a willingness to treat the supply chain as a strategic asset that deserves the same consideration a media company brings to content acquisition or audience development. The companies that make that investment tend to compound. The ones that don’t will be absorbed by the ones that did, at valuations that assume the leaking never gets fixed.

That is the debt penalty. It has been coming due for a while. It is due now.

Metadata determines whether content reaches the right audience, whether rights obligations are met, and whether revenue is captured or lost. It is foundational to how media businesses operate. When treated as a technical afterthought, the business pays in failed launches, missed revenue windows, and compounding operational drag. This piece makes the case for treating metadata as an executive priority.

Metadata Belongs on the Executive Agenda

Every major business decision in media relies on metadata: what content is available, where it can be shown, who holds the rights, how it is priced, whether it meets compliance requirements. These variables determine whether a business executes on its strategy.

The failures that follow from poor metadata rarely announce themselves as a metadata problem. A content launch stalls because rights windows are unclear. An ad sales team loses a deal because audience segmentation data is unreliable. A FAST channel drops out of a platform because an ingestion schema is misaligned. In each case, someone is managing a workaround while the root cause goes unaddressed.

If your metadata is not aligned with your business goals, it is working against them.

Case Study: Pluto TV and the Speed of Alignment

I joined Pluto TV as employee 42. Before the Viacom acquisition, we faced a problem that will sound familiar: metadata was fragmented across teams, tools, and systems. Content delivery was slow. Ad targeting was inconsistent. Compliance was reactive.

We finally got approved for a new MAM, and a precursor to technology is a working data model, so we gave ourselves eight weeks to fix ours across a content library sourced from 200 different providers. We did it. We aligned metadata across business, product, engineering, data, and operations, and the workflows accelerated, ad performance improved, and compliance moved from reactive to proactive.

What happened next is the part of the story worth dwelling on. After the Viacom acquisition, we watched their teams spend over a year attempting to replicate what we had done in two months. The technical leaders on their side were smart, collaborative, and capable. The challenge was structural. Their environment was not built to support strategic metadata alignment at scale.

Technical complexity was not the differentiating variable. Organizational alignment was.

What Made the Pluto TV Approach Work

Three things mattered:

Clear Business Goals. Everyone understood what we were solving for: faster content delivery, stronger monetization, and airtight compliance. Every decision was grounded in those goals. The teams that get stuck on metadata work are usually optimizing for process completion instead of outcomes.

Cross-Functional Ownership. Metadata stopped being a technical responsibility owned by one team. Every function had a seat at the table to define the business rules for the metadata they relied on. That changed who was accountable, which changed what got prioritized.

Outcome Accountability. Progress was measured by business impact, not by system updates. Schema validity is a floor. The ceiling is whether metadata is delivering against the revenue, rights, and compliance outcomes the business actually cares about.

The Results

Distribution workflows accelerated. Ad inventory became more accurate and easier to sell. Rights logic was consistently applied. Compliance moved from reactive to proactive.

None of this was about headcount or technology. It was about strategic focus and execution.

When Metadata is Misaligned, Strategy Suffers

Revenue is left on the table because ad targeting metadata is incomplete. Content investments underperform because discoverability metadata is inaccurate. Rights violations create liability. Compliance failures require expensive retroactive remediation.

These signal a deeper gap in business strategy. Metadata misalignment is almost always a symptom of organizational fragmentation: siloed teams, no shared standard for what metadata means across functions, and no executive ownership of it as a business asset.

If metadata does not reflect how your company earns, protects, and delivers value, every other system is forced into expensive workarounds.

Schema Alone is Not Enough

A well-designed schema is necessary but insufficient. I have seen organizations invest heavily in schema governance that fails to translate into business performance: the people filling the schema do not understand why the fields matter, and there is no feedback loop between metadata quality and outcomes.

Schema tells you what metadata should look like. Strategy tells you what it needs to do. That gap is where operational failures live.

Interoperability is a Strategic Imperative

Content moves across the industry. At every handoff it must carry consistent rights information, consistent content identification, and consistent technical parameters. When it does not, every party absorbs friction.

Interoperability is a structural challenge requiring industry-wide alignment. Organizations that participate in standards-setting work protect their own operational efficiency and build the shared vocabulary that makes faster integration possible.

Questions You Should Be Asking

Five questions worth bringing into your next operational review:

1. Who owns metadata strategy, with authority across functions?

2. How long does it take to onboard a new distribution partner, and what causes the delay?

3. Where are we absorbing operational costs because of metadata inconsistency or remediation?

4. How do we measure metadata quality, and is that measurement tied to business outcomes?

5. Are we participating in industry standards work, or waiting for others to set the terms?

These are strategic levers. Treating them as IT hygiene is how organizations end up paying for misalignment.

Why This Work Matters

When metadata reflects how a business actually operates, the returns compound. Content reaches the right audience. Ad inventory is priced and delivered accurately. Rights are enforced without manual intervention. Compliance is built in, not retrofitted.

When metadata is mismanaged, every system downstream compensates. That shows up in slow launches, lost deals, and teams working on rework that should not have been necessary. It is costing you. Consistently. Across every part of the operation.

Metadata is Executive Strategy

Metadata is strategic infrastructure. It belongs in the executive conversation.

The organizations that understand this treat metadata the same way they treat financial data or legal compliance: as a business asset requiring clear ownership, defined standards, and accountability to outcomes.

The organizations that do not are managing symptoms. They fund workarounds, absorb avoidable delays, and leave revenue on the table.

Metadata is a business decision that shapes whether everything else works. It belongs on the executive agenda because that is where the leverage is.

That sentence took a while to write. I’ve spent the last two years betting on an idea I believed in completely, and saying it out loud means admitting the bet didn’t pay off. I’m not someone who fails gracefully. I don’t have a lot of practice with it. This one was monumental.

So let me say it plainly: I failed at building a sustainable consulting practice in this market at this moment. I want to be precise about what failed, because precision matters here. The capabilities didn’t fail. The work product didn’t fail. The insight didn’t fail. The impact I had with clients didn’t fail. The writing, the brand, the framework I built and gave away freely: none of that failed. What failed was the market thesis: that the companies I was trying to serve were ready to pay for someone to come in and take ownership of their operations with them. They weren’t. And I did not see that clearly enough before I started.

I’m writing this because building something on your own is extremely lonely work. I believe personal connections rule everything, and I want to write the thing a lot of people don’t want to talk about, in case there is someone reading this who can relate. If that’s you, we are of the same tribe.

Why I Built This in the First Place

I launched Integration Therapy because I had a particular kind of vantage point that I had not seen many other practitioners bring to the work. I had been inside all of it.

Zero-to-one streaming launches. Networks that never got past zero-to-one. Networks that blew up into enormous successes. Enterprise broadcast companies in the middle of transforming into streaming. Different scales, different cultures, different failure modes, different architectural decisions, and a clear view of what marked the difference between the companies that scaled cleanly and the ones that scaled into chaos.

That range taught me something I do not think you can learn from any single seat: the supply chain and the value chain are one system. The companies that treat them that way win. The ones that don’t bleed out slowly and wonder why their margins keep compressing.

I also went back and got a master’s in organizational management and leadership while I was building this because operating that range of companies made one thing obvious: teams are central to transformation, and most operational problems are organizational problems wearing technical costumes. I wanted to be able to name what I was seeing more precisely, and I wanted the academic framework to back up the operational experience.

Working across that many cultures gave me something else, too. I learned to recognize a healthy one quickly, and an unhealthy one even faster. More importantly, I am still learning to understand the systemic failures that produce bad cultures in the first place. That work has shaped how I think about operations as much as any individual engagement, because the cultural layer is where the operational architecture either holds or breaks. Especially right now, in an industry that is fragile from the AI shift, from the constant layoff cycle, and from the loss of institutional knowledge that has walked out the door over the last three years.

By the time I left my last operating role, I had seen the same pattern in enough companies to know it was systemic. Operational complexity accumulating into real financial drag. Teams doing heroic manual work to compensate for process gaps no one had been given permission to fix. Metadata decisions made by vendors, by default, because no one had stepped back far enough to see the architecture. Millions of dollars sitting in the gap between what the operation was capable of and what it was actually delivering.

I thought: I can help fix this. I built the frameworks. I built the diagnostic tools. I built an AI stack that let me deliver the kind of analysis that used to take months in a matter of days. I wrote 32 articles explaining exactly how I think about the problem. I gave the methodology away. I spoke at conferences. I chaired the Metadata Working Group at the SVTA. I showed my work, publicly, because I believed that showing the work was the fastest path to finding the companies that needed it.

Some of them found me. The client work I did was genuinely rewarding. Every engagement where I had the chance to go deep and make a real impact reminded me exactly why I started.

But I couldn’t make the model sustainable. And I’ve spent a lot of time since then trying to understand why.

What I Got Wrong

The honest answer is that I misread the market in a few specific ways.

The first: Companies see their operations as a pain point, but rarely as an urgent one. The dysfunction is real, and most of the operators inside these companies know it intimately. But the executives who sign consulting agreements have learned to live with the pain. They’ve normalized it. It doesn’t feel like a crisis; it feels like Tuesday. And that gap, between the people who feel the problem every day and the people who have the authority to commission a solution, is where a lot of consulting engagements go to die.

The second: A significant number of streaming companies are content to let their vendors make operational decisions for them. I’ve thought a lot about why. Some of it is change aversion: the perceived risk of redesigning something that mostly works feels higher than the cost of running it at 70% efficiency indefinitely. Some of it is that the strategic value of owning your own data layer isn’t obvious until something catastrophic exposes the gap. And some of it is something I still don’t fully understand. There’s a version of this story where learned helplessness is the answer. There’s a version where the incentive structures inside these companies make it genuinely rational to outsource the thinking. I don’t have a clean conclusion. But I think about it constantly.

The third: All the industry conversation about operational efficiency, about AI readiness, about metadata as a strategic asset: I thought that conversation was moving toward action. The way the discourse read, companies were waking up to the idea that their operations were a competitive lever, not just a cost center. What actually happened is that the technology companies stepped in and sold the next wave of solutions. The promise was that the right platform would make operations simple. Buy this tool, eliminate the complexity. I’ve been inside enough organizations to know that’s not how it works. Technology doesn’t simplify operations. Operations that are well-designed can be supported and accelerated by technology. That’s a different proposition. But it’s a harder sell, and the vendors have larger sales teams.

The fourth: The short duration of consulting gigs is a structural problem for this kind of work. The impact I made with clients was real. The results were measurable. But getting to the next engagement requires a constant hustle that, over time, pulls attention away from doing the work and toward finding the work. That cycle is exhausting, and it’s incompatible with the kind of deep operator engagement I do best.

The fifth: I have a strong point of view about how streaming media companies should architect their supply chains, their contracts operations, and their metadata operations. That point of view is built on real experience across multiple companies at multiple scales, and I’m not willing to soften it to close a deal. The market, right now, is not consistently willing to pay for a practitioner who comes with opinions. It wants someone to confirm what it already believes, or to implement what’s already been decided. I’m not that person.

The Vendor Question I’m Leaving Open

I want to say one more thing about the vendor dynamic, because I think it’s the most important unresolved question in this industry.

The companies that build MAM systems, CMS platforms, OVP tools have become, by default, the people who decide how streaming operations work. Not because they set out to own that role, but because someone had to fill it, and they were in the room. When a company buys a platform and then lets the platform vendor configure their workflows, define their metadata schema, and determine what their operational model looks like, they have handed over something that is genuinely strategic. The vendor’s incentive is not to build you an operation that makes you less dependent on them. In the age of AI, where the quality and structure of your data determines what you can actually do with it, that dependency has compounding consequences.

I don’t have a clean resolution to that pattern. I’m not sure anyone does yet. But I think the companies that figure it out first are going to have an operational advantage that’s very difficult to close from behind. I’m leaving that thread open here because I think it deserves more conversation than a paragraph, and because I’m genuinely curious what others are seeing. If you’re seeing it differently, I want to know.

What I’m Proud Of

I took the big swing. I built something real. I did it on my own terms, with my own money, with a framework I developed from two decades of operator experience. I published work I’m genuinely proud of. I had conversations with some of the most interesting people in this industry about the things that matter most to the work. I built an AI stack that let me produce diagnostics and communication strategies at a speed that would have been impossible five years ago. I gave a lot away for free, and I know some of it landed.

The clients I served and the places where I made a real impact were extremely rewarding. The writing was rewarding. I built a brand with a point of view, and that matters to me.

And on top of the work I have been able to do, the framework I have been able to build, and the content I have been able to create in the last two years, I will never be able to unlearn the things I have learned sitting with so many executives at so many media companies who were willing to be candid about what isn’t working for them. Those conversations changed how I think. Those relationships are ones I will always value.

And I am extremely proud of myself for having done this. I don’t like to fail. This one stings. But I would rather have taken the swing and missed than have spent two more years inside someone else’s broken system wondering if I could have built something of my own. The answer is: I could. I did. It just didn’t work.

What’s Next

I’ve converted the Integration Therapy website to a Substack. I’m not going to stop writing. The thinking that drove the business doesn’t disappear because the business is closing, and some of my best work has come from having a platform where I could work through ideas in public. I’m hoping to reach more people there and have more collaborative conversations. If you’ve been reading, I’d love it if you kept reading.

On the professional front: I am looking for a full-time role.

I want to be specific about what I’m looking for. I want to work with a company that wants to innovate and leverage every asset it already has in-house. Where people, process, and technology are designed to work together across the entire value chain, not optimized in isolation and then bolted together. Where senior operations leadership isn’t just a function that keeps the lights on, but a genuine strategic input.

I am a zero-to-one operator, a scale operator, and a transformation operator. I’ve been all three at different points in my career, sometimes at the same company in the same year. What I’m looking for is a company that wants to deploy that full range: not in isolated consulting engagements, but with the continuity and institutional depth that only comes from being inside. The kind of role where I get to bring everything I know to one place and build something that lasts.

If that sounds like your company, I’d like to talk.

A Note to Everyone Who’s Read, Responded, Referred, and Engaged

You know who you are. The executives who gave me time when you didn’t have to. The operators who told me I was right about things they couldn’t say out loud inside their organizations. The peers who collaborated, challenged, and pushed the thinking. The clients who trusted me with real problems and gave me the room to solve them.

This industry has genuinely interesting people in it. That part I wouldn’t trade.

---

All you have to do is turn on your TV at the end of the night, watch anything in streaming, and use your imagination to see how much green pasture there is out there. I’m really excited to see what the next companies do. And I cannot wait to be a part of one.

Parks Associates released a report with Philo showing that the average U.S. household now subscribes to 5.3 streaming services. More than 300 streaming services are competing for that wallet share in the U.S. alone. And 33% of pay-TV subscribers say they stay with their provider specifically because they can find more of the content they want in one place. That last number is the one networks are paying close attention to.

Fragmentation fatigue is real. Consumers are tired of hunting across several apps to find one thing to watch. Aggregation is becoming a competitive advantage, and content is increasingly appearing on more than one platform at a time, sometimes several. Companies are competing on discovery, on curation, on depth of catalog. That pressure has driven a wave of catalog acquisitions as streamers race to own more content and put it in front of more people, faster.

The strategy is sound. The execution is where it falls apart. Major catalog acquisitions include acquisitions of content metadata, and content metadata is largely non-standard across the industry. This includes rights data, which can create an enormous amount of avoidable operational drag when it is not handled mindfully.

The Rights Data Does Not Translate

Here is what is actually happening inside a lot of streaming companies right now: They have catalogs. Big ones. Sometimes acquired from other companies, sometimes licensed libraries, sometimes legacy content sitting in a system for years under contract terms that have not been fully translated into today’s rights management systems. And when they go to launch a FAST channel, expand into a new territory, or add a content package to an OTT offering, they hit a wall.

Many networks cannot figure out where they are allowed to show large swaths of their content.

From my experience, this is largely a tier one, tier two streamer problem. It exists inside companies with massive infrastructure, headcount, and revenue. Although operational complexity looks different from company to company, the symptoms feel consistent. Contracts were interpreted one way in one system but differently in another. Rights data was interpreted a certain way for one catalog but not that way for another. Does “United States” include the territories, or just the states? Does it include Guam? ISO-3166 codes were used across all of the libraries, but it turns out the acquisitions SVP at the old company used them with rather loose vernacular definitions.

Compliance has to be tuned tightly to the appetite for risk of the organization. If you don’t want to get sued and you don’t want bad publicity, the data needs to be accurate. And even if you are comfortable taking the legal risk, you don’t necessarily want to risk a relationship with a content partner that might have great content down the road that you are going to want.

The business wants to move fast, and many rights operations cannot keep up right now, which is a very expensive latency because content has already been acquired cannot be monetized. There are larger tier two companies today that are unable to launch the number of channels they want, and therefore unable to generate the ad inventory and revenue they need, because the people managing their supply chain are having a very difficult and very slow time figuring out what can actually get distributed where.

The Vendor Instinct (Check Yourself Before You Wreck Yourself)

When companies feel this pain, the instinct is to buy software to fix it. What tends to happen is companies get overwhelmed by how complicated this data problem is and start classifying it as just a data problem, or just a technical problem. It is an operations problem.

There are real players in this space: Rightsline, FilmTrack, FADEL, MetaComet. These are legitimate, capable tools. The market is competitive enough that each has developed its own approach, its own data model, and its own workflow logic. Some of them are genuinely excellent.

But they are tools. They know how their own software works. They do not know how your company works. They do not know who depends on your rights data, in what format, or at what point in your supply chain. They do not know your programming strategy, how your distribution team makes its decisions, or how compliance is handled inside your organization. They will not learn it as well as you know it. This is not a criticism of these vendors. It is simply true, and it is not fair to expect otherwise.

Companies are treating rights management platforms as if they are rights management strategies. They are different things. Keep your strategy in-house. You can outsource implementation. But the strategy has to be yours. When you rely on a rights platform vendor to solve your whole rights operation, that vendor will be cornered into retrofitting your company into their model, and they’re often not even aware of it.

When you put a vendor on top of an undefined rights operation, here is what happens. You had a spreadsheet problem. Instead of organizing the spreadsheet, instead of defining your terms, instead of connecting the data to what your company is actually trying to accomplish, you bought software. Now you have a spreadsheet problem and an implementation problem, and the vendor is charging you for both. The underlying mess has not been cleaned up. It has been digitized. And the tool will faithfully automate whatever chaos it finds. This is a great example of avoidable operational complexity, where you’re moving fast, but you’re realizing less revenue, and more of the revenue you’re realizing is leaking out before it hits the profit line.

It is a lonely problem to have inside your company, but if you feel this pain, you are not alone. A company acquires a catalog. The acquisition comes with rights data maintained in a system built for a completely different operating model. The new owner plugs it into their rights platform and proceeds. Six months later, the team is manually overriding the system on a Google doc because the logic no longer holds and there isn’t roadmap or budget space to call the vendor back in so soon. The rights platform is doing exactly what it was configured to do, but there are too many caveats that it cannot hold.

The configuration was actually wrong from day one, because nobody defined what the rights operation should look like end-to-end before the implementation began. And this goes well beyond deciding whether to use ISO-3166. Legal compliance needs to match your contracts, which need to match your supply chain, all the way through.

What You Need to Define Before You Evaluate Vendors

A rights strategy is not a rights management platform. It is the layer underneath the platform, and it has to exist before you bring in the technology, because it is the business requirements for your RFP and your vendor.

Before you start evaluating vendors, your organization needs answers to a set of foundational questions.

What is your company trying to achieve with its content? Not in abstract brand terms. In distribution terms. Which platforms, which territories, which windows, which revenue models? What does a tight rights clearance workflow look like for your organization right now? What about marketing materials? Who is accountable for all of this?

How do you actually use the data? Who touches it? Who depends on it to make choices that move the company forward? At what point in the supply chain and in what format? Is your programming team making scheduling decisions based on ratings, or based on content strategy and then requesting the rights? Is your metadata team flagging territorial restrictions at ingest? What are the business rules that surround that? Is your distribution team running clearance checks before delivery? How does that work? If those teams are working off different versions of the same information, you have a structural problem that will only get worse the more horsepower you put behind it through vendor automation or AI. Software does not fix structural problems. It inherits them and makes them bigger.

What does compliance look like in your operation? This question reveals more than any other. Companies that have not defined their compliance framework end up with rights platforms that are technically sophisticated and operationally useless, because nobody can agree on what a compliant distribution decision looks like.

Once you have answered those questions, you have a rights strategy. Now you can evaluate vendors. The question is which of these tools fits closest to the data model you need to support and the strategy you are trying to execute. Do not use who is familiar to you. Use who works best with your company.

Rights and Attributes: What Are They, Who Needs Them, and How Are They Used?

There is another layer that companies need to define before they bring in any technology: the distinction between rights and attributes. In the data hierarchy, rights are the parent. Attributes are the child. Rights define what you are allowed to do with a piece of content. Attributes define the conditions under which that content gets delivered to a specific user on a specific device. Both are essential these days. We used to operate primarily on rights alone, but attributes have become much more central over the last three or four years as streaming audiences have exploded across devices and started consuming content in different ways.

Rights data at a company generally means the business level. Territory. Channel inclusion and exclusion. Windowing. Frequency caps on FAST channels. Time-based restrictions. These govern when and where your content can appear.

Attributes are more specific and operate further downstream, at the device and user level. What devices can a title stream to? Can you go anywhere except Samsung? How many concurrent plays are allowed in the same household? How many total plays per month can an OTT title have per user? This layer governs how content gets consumed once it has been cleared for distribution.

Every company I have worked with defines the line between these two things a little differently depending on content type, content partners, and distribution strategy. Because rights and attributes are often handled by different systems at different points in the supply chain, the question of where this data lives matters enormously. At some companies, the business-level rights data lives in Rightsline or FilmTrack, and the attribute data is handled further downstream by the product team.

I do not recommend that separation. All of this data should be easy to access during reporting, and ideally manageable from the beginning of the supply chain. This is something I have seen mostly at tier one companies where the product team handles attributes without direct access to the larger rights data. There are booby traps in that arrangement.

A holistic view should be available to business teams, operations teams, and product teams. When people making decisions about partner-facing, compliance-critical data do not have a complete picture of how rights and attributes connect, you are setting yourself up for very big, very urgent problems.

Define what rights are. Define what attributes are. Figure out who benefits most from seeing all of it. Consider putting it all in your rights management system from the beginning. Measure twice, cut once.

This Work Should Come Before Catalog Ingest

Catalog acquisitions are happening at scale, while programming decisions are happening with white gloves. Companies are acquiring content libraries to compete on depth, fill out FAST channel grids, and anchor subscription value propositions.

Every catalog acquisition brings rights data maintained under somebody else’s operating model. The contracts exist. The underlying data is probably there. But it should be checked for completeness, accuracy, and how it differs from the acquiring company’s approach to rights and attributes. That often does not happen. People rush to get the catalog into their systems thinking they will figure it out later. Now they have a gigantic data mess.

That data was interpreted in another company’s system, with their definitions, their values, their appetite for risk, under their own compliance logic, to serve a different distribution strategy. That interpretation does not automatically port over.

When you have not defined your own rights operation before catalog ingestion, you are adding an enormously complex data set on top of an undefined system. You are also leaving revenue on the table, because that content is going to sit in the library while people untangle the data.

You cannot implement your way out of a strategy gap.

Ideally, the work looks like this: establish how your company interprets and stores rights data, define your compliance requirements, map the downstream dependencies, choose and implement the vendor that fits. Then when the catalog arrives, you have a system that knows how to receive it.

I also want to be realistic. The idea that you will have a perfectly defined rights strategy and a vendor implementation complete before the catalog shows up is not usually how it works. Acquisitions move on their own timelines. But here is what you can do: resist the urge to converge everything at once.

Think of it as three separate tracks. Your existing catalog and rights data. Your new rights management system. The newly acquired catalog. Do not combine those tracks until each one is clean, defined, and ready to fit together. It should look like a Tetris board, not spaghetti. The urge to combine everything at once usually comes from a sense of urgency to go to market as fast as possible. That urge is what slows you down, because converging three messy tracks into one big messy track does not get you to market faster. It gets you to a standstill.

Conclusion

Rights management platforms will continue to improve. Competition in that space is good for everyone. But the platform is the last piece, not the first. The first piece is understanding what your company is actually trying to do with its content, how the data needs to flow to support that, and who needs to see what along the way.

If you are reading this and recognizing your own operation, you are not behind. This is where most of the industry is right now, from my experience. The companies that pull ahead will be the ones that take the time to define their rights strategy before they let the technology define it for them. That work is not glamorous and it does not move fast. But it is the work that makes everything else move faster once it is done.

This is the third piece in a series I didn’t plan to write as a series. The first, The Organizational Trauma of Layoffs, was about naming the damage: what actually happens to the people and systems inside a company when a layoff hits, and why the industry keeps pretending it doesn’t. Layoffs aren’t my specialty. I’m an organizational and media operations expert. But the unplanned, wild success of that article, and how often it gets googled by people in this industry and beyond, inspired me to write a follow-up for my Streaming Wars column, Myths in Streaming. That entry, What Are Middle Managers Even For?, looked at a specific version of that mistake: streaming companies cutting the people who hold operations together and calling it “organizational agility.” The promise was that decision-making would get faster. It actually got slower, and now many of those positions are starting to slowly get hired back.

This third article answers another question I’ve been asked after both of those: OK, fine. So what do we do about it? Which is actually where my real lens comes in. I work with streaming companies that are hemorrhaging revenue before it reaches the profit line, helping them optimize their operations and build frameworks that allow them to consistently do more with less without sacrificing quality. Unfortunately, this often means I’m coming in right before or right after a round of layoffs.

The timing matters. CBS News cut 6% of its workforce in March 2026. WME cut 3% the same week. Lionsgate, Axios, Universal Music Group, and Netflix’s global product team all announced cuts within days of each other.

This comes after more than 17,000 entertainment, news, streaming, and broadcast jobs vanished in 2025, an 18% spike from 2024’s already punishing numbers. Consolidation is increasingly cited as the driver. The Paramount/Skydance merger alone is expected to produce another wave, on top of the damage that’s already been done. And let’s not forget the Warner Bros. Discovery acquisition of Paramount, which will compound consolidation cuts further still. The industry is doing this on a loop.

We keep treating the symptom as the treatment. And it isn’t working.

The Reframe

Before I get to the playbook, I want to be clear about what I’m not arguing. I am not anti-layoff. That framing is too simple, and I’ve been inside too many companies to pretend that headcount decisions are always avoidable. They’re not. Markets shift. Technology changes. Audience habits evolve faster than most organizations can adapt. Sometimes a company genuinely needs fewer people than it has. Denying that is its own kind of dishonesty.

Companies need to innovate. They need to evolve, to grow, to have some appetite for risk. And every once in a while, failures happen. The idea of having to lay people off to save the company is not, in itself, criminal.

But there are three things I think most leaders get wrong when layoffs happen, and getting them wrong makes everything so much worse.

Principle 1: A layoff is not a strategy.

It is an admission that something failed. A bet that didn’t pay off. A model that didn’t scale. A market that moved faster than the plan. There is no shame in that. Failure at scale is often just the cost of swinging for something real. But there IS shame in dressing it up as “optimization” or “right-sizing,” as if the company identified an exciting new efficiency rather than hitting a wall.

The first honest thing a leadership team can do after a layoff is name what actually happened. Not in HR language. In plain English. This failed. Here’s what we learned. Here’s what we’re changing.

That kind of clarity is not a weakness. It is the only foundation you can build anything on. And lies are often found out, especially in an age where information exists literally everywhere. Consider how many of the layoffs in 2024 and 2025 were blamed on AI. We now know that almost none of them were directly caused by AI. Companies are actually struggling to implement AI because their operating layers are messy. And in many cases, their operating layers are messy because they were hasty to lay off. When you blame something that isn’t the real reason, people find you out, and your reputation doesn’t easily recover. Maybe that’s acceptable when you’re a CEO with a $15 million parachute in your contract. But most of the people who get shoved in front of the teams to deliver the news are not that.

Principle 2: The language matters more than leaders think.

“Affected by layoffs” is a phrase that makes my skin crawl every time I see it. In a private email. On a client’s desk. On social media or in articles. Anywhere. It needs to be destroyed. Everyone in the building is affected by layoffs. So are their families. The people who lost their jobs were laid off. Say that. Use that word. It is accurate, and it is a matter of basic humanistic integrity to name the damage you are causing. It is deeply disrespectful to take away somebody’s job, turn their life completely upside down, and then not even say what you did.

Corporate euphemism isn’t compassion. It’s avoidance. And everyone still in the building can tell the difference. When leadership hides behind carefully constructed language, the message the remaining team actually receives is: this is how honest they’ll be with us going forward. That message sits. And it shapes every conversation about the company’s future from that point on.

The leaders who have a real shot at rebuilding swiftly after a layoff are the ones who show up with courage. Who stand in front of their teams and say, plainly and without a PR filter: we failed here. That is not a legal script. It is a choice. And most leaders don’t make it, because it requires them to be accountable in a way that feels personally exposing. I understand that impulse. It is the wrong impulse. What you have done to the people who were laid off, the people who remain, and the families of both is deeply personal. That is why leaders get paid the big bucks, to be crass. Show up. Don’t use legal exposure as an excuse for cowardice. I’m saying this as a person who has been on both sides of the table.

Principle 3: Trust is the real casualty, and rebuilding it is the real work.

A layoff breaks the psychological contract between employer and employee. That’s not a metaphor. It is a measurable collapse in engagement, loyalty, and willingness to take risks. Research on organizational trauma consistently shows that the damage doesn’t stop with the people who leave. The people who stay are often more destabilized, not less. More overworked, not less. Survivor’s guilt is real. Anxiety about the next round of layoffs is real. Distrust of leadership is real, and it persists long after the stock price recovers. This leads to fear of innovation and a lagging bottom line.

You cannot fix that with a town hall. You cannot fix it with a pizza party or an all-hands where the CEO says “we’re through the hard part.” You rebuild trust with time, consistency, accountability, and a real internal communication strategy that feels transparent. By showing the team, repeatedly and over months, that leadership has actually learned something, changed something, and won’t pretend the damage didn’t happen.

The companies that recover fastest are not the ones that immediatly galvonize and try to just move on. They are the ones that slow down long enough to be honest about what happened. They let the grief process happen. And then do the slow, unglamorous work of demonstrating that they care about their teams.

The Playbook

If you are a leader inside one of these organizations right now, whether you just made cuts or you’re anticipating them, these are the four moves that actually matter.

Move 1: Diagnose before you cut.

Most companies don’t do this. They cut by department percentage, or by seniority band, or by how long somebody has been there, or by what their job description says on paper. When was the last time anybody’s job description sitting in their HR folder was actually accurate? They do it anyway. They remove a percentage and hope the remaining structure holds.

Before any headcount decision, map the operating model, not what’s on paper. Follow how decisions get made all the way through your value chain. Where does work actually happen? Who owns what? Where are the dependencies? What specifically breaks if you remove a given role? And understand who possesses the most institutional knowledge, and why that knowledge is stuck in the head of one person. That question never gets discussed, and it absolutely needs to be. Is that person a cornerstone of your organization, or are they a bottleneck? Figure it out. Do the work. Measure twice, cut once.

This isn’t an academic exercise. It’s the difference between a cut that tightens an organization and a cut that destroys the capacity to execute. None of this damage is “quiet” or surprising. It didn’t sneak up on anyone. People weren’t paying attention.

In my executive roles, I’ve been on both sides of this table. As an advisor, I’ve done operational diagnostics before headcount decisions, and I know that work saves millions in downstream damage. Without it, everything else is guesswork. How much time do you have for wrong guesses?

Move 2: Redesign the work, not just the org chart.

This is where most post-layoff planning fails. Leadership reduces headcount, redraws the boxes on the org chart, and expects the work to continue at the same volume and quality with fewer people. It doesn’t. It can’t. This is physics.

If you’re going to operate with 20% fewer people, the work has to change. Can you still operate with 20% fewer people? Hopefully you knew that before you made the cuts. But probably so, yes. Which workflows need to be consolidated? Which approval layers are actually adding value versus adding latency? Which manual processes can finally be prioritized on your technical roadmap to be automated for real?

The goal is a right-sized operating model. Not a shrunken version of the one that already wasn’t working.

Contraction is not just an opportunity to redesign. It means that redesign is critical. And most operating models need to be completely rethought, not slightly retooled. This is an opportunity to get lean, to get competitive, and to do it while leaking a lot less revenue.

Instead, most companies spend the period after a layoff putting out the fires that the layoffs themselves caused. They waste the opportunity to redesign the entire operating model.

Move 3: Protect the connective tissue.

Middle managers, senior individual contributors, and cross-functional coordinators are reliably the first to go in a cost reduction. They’re expensive, and it’s difficult to quantify their value. Often they do jobs that other people simply don’t understand, because other people belong to a vertical and these people, by definition, do not. But these are the people who can see things that nobody else can see. It’s a real shame that they automatically get cut.

I wrote about this at length for The Streaming Wars in January, but I’ll say it again here: these are the people who make the system work. They translate strategy into execution. They catch errors before they compound. They hold institutional knowledge that took years to build and can’t be documented in an offboarding call. When you cut them, you don’t get a leaner organization. You get a flatter one where decisions bottleneck at the top. And you get a more siloed one, splitting the company into a bunch of little whack-a-mole organizations that can’t coordinate across boundaries. When you start cutting cross-functional people, you need to understand where they’re adding value in a real way. You have to be extra thoughtful about that.

If you must reduce, protect the roles that connect systems, teams, and decisions. Cut the roles that duplicate effort, not the ones that prevent chaos. And make sure that when you say you’re building leaner decision processes, you are actually following the lifecycle of how decisions get made, rather than just cutting middle management and leaning on the company line.

Move 4: Build the operating model you’ll need in 12 months, not the one you’re surviving today.

Contraction might be temporary. The operating model you build during contraction is what you’ll be running on when growth returns. And it will return. It always does, eventually, in some form. Whether that form involves human beings, AI, automation, vendors, consultants, or partners, if you are trying to grow your company in any direction, you are going to have to grow your operation.

Companies that cut without redesigning end up hiring back into the same broken structure about 18 months later. Different job titles, same dysfunction. It’s a tax, paid in turnover, retraining costs, and corporate reputation. Every time a company goes through this cycle, it ends up hiring worse people, because the best people are going to companies that don’t do this and staying there. It is really hard to recover your corporate reputation when you keep acting like this.

Use contraction as the forcing function to build the model you should have built during the growth phase, when you had the money and the momentum and somehow never found the time.

For the People Left in the Building

I want to shift gears and talk directly to the operators, managers, and ICs who survived the last round and are now doing more work with fewer people, carrying guilt they didn’t earn and fear they didn’t ask for. I’ve been that person. This part is for you.

1. Name what you’re feeling. It’s not dramatic. It’s accurate.

Survivor’s guilt is real. So is grief for colleagues who are gone, anxiety about the next round, and a deep distrust of anything leadership says right now. These are normal responses to an abnormal event. You are not overreacting, and you are not being disloyal by having them. The company may be signaling that everything is fine now, that it’s time to move forward. It may not be fine. You are allowed to know that, and you are allowed to not be fine. You might have a family. You have a life outside this building. The ripple effect of what just happened extends well beyond your desk, and that weight is real too.

2. Get clear on what you can and can’t control.

You cannot control whether there’s another round. You cannot control the strategy, or make leadership be honest if they won’t be, or fix a broken operating model from the middle of it. What you can control is how you show up for your team, what standards you hold for your own work, whether you let fear flatten your judgment, and whether you keep investing in your skills and your network. Agency isn’t about controlling the outcome. It’s about deciding how you move through it.

A note for the mid-level managers and directors who are trying to make their teams feel okay: your teams are going to be hurt. They’re going to feel betrayed. They’re going to be going through grief. And you cannot really do anything about that. What you can do is take care of yourself the best way you can, show up as fresh-faced every day as you possibly can, and make space for people to feel frustrated, to express themselves, and to grieve. You can control whether or not you’re galvanizing your team too soon. You can control whether or not you’re letting them have a voice in what changes moving forward. But you cannot control how other people feel. I need people to hear that, because so many folks in mid-management are incredibly empathetic, and they go under an enormous amount of stress when their team is hit with layoffs.

3. Document everything. For yourself.

When an organization is in chaos, the people who stay clear-eyed about what’s actually happening are the ones who come out of it stronger. Keep a running log. What are you working on? What problems are you solving? How are those problems tracking up to corporate KPIs? What are you learning? What are you building? Not for your performance review. For you. So that when the dust settles, you have a clear record of what you survived and what you’re worth. Not just to that company, but to the industry and to yourself.

4. Protect your limits. The work will expand to fill every hour if you allow it.

After a layoff, the implicit expectation is that everyone works harder to compensate. That expectation is the company’s problem cosplaying as your responsibility. You can be excellent and still have limits. In fact, the best operators I work with know and understand that burning out doesn’t make you loyal. It makes you less effective, less present for your team, and less employable. You don’t owe this company your health. You don’t owe this company your sleep. You don’t owe this company your well-being. And what you really don’t owe this company is doing a bunch of invisible work that they are not even aware of. It creates giant holes in the system that they cannot see and fill the right way when you are constantly compensating for their shortcomings in strategy. Sometimes, advacating for yourself is the same thing as advacating for the long-term health of your employer.

5. Don’t stop building your external network.

The people who feel most trapped after a layoff are the ones whose entire professional identity lived inside one organization. Stay visible. Stay connected to people outside your company. Your loyalty should be to your career, not your company. Your next opportunity, whether it’s inside this company or somewhere else, will come from the relationships you maintained when things were hard. They also come from the relationships you maintained when things were easy. Not the ones you let go because you were exhausted or uncertain or just didn’t want to reach out.

The through-line for all five of these: you did not create this situation. But you do get to decide how you move through it. Employees have a lot more agency than they are generally told. The people who come out of these moments strongest are the ones who refused to let someone else’s shortcomings define their trajectory.

The Close

The streaming and media industry has normalized layoffs as a management tool. So has tech. The numbers make that impossible to dispute: 17,000-plus jobs gone in 2025, more cuts already in the first quarter of 2026, and consolidation still accelerating. Cutting headcount has become the default response to pressure, the move that buys time and signals decisiveness to Wall Street, regardless of whether it actually addresses what’s broken.

But a layoff is not a plan. It is not a minor correction, and it is not a strategy. It is a reckoning. And what comes after it determines everything.

After working with media companies before, during, and after layoffs, I can say from experience that the companies that will emerge from this era in the strongest position are not the ones that cut the deepest or moved fastest to a town hall. They’re the ones that used contraction to get operationally honest with themselves, with their employees, and with their shareholders about how their business actually works, where the inefficiencies live, what they need to build for the next phase, and how to make themselves more resilient.

The question I started with, what do we do about it, has a less satisfying answer than most people want. You do the diagnosis. You redesign the work. You protect the people who hold the system together. You tell the truth about what happened, plainly, without the PR filter. And you rebuild trust one consistent action at a time, because there is no shortcut for that part, and because your best workers are the ones who can most clearly see whether or not you are acting with integrity.

The companies that are willing to do this work are unfortunately getting rarer. But they are also the ones worth building. And from sitting in their offices, I know they are the ones that will win.

In the legacy broadcast world, compliance was a manual, frame-by-frame endeavor managed by massive “Standards and Practices” (S&P) departments, which were rooms full of people whose entire job was to watch every second of video to ensure it met legal, regulatory, and advertiser standards.

In the high-volume streaming world of 2026, that model is an operational suicide mission.

If you are a streaming distribution network today, your volume of content is way, way too high to manually review. It is, in fact, too high to pragmatically or diligently check even using AI. If you are operating a lean team, you cannot conduct yourself like a legacy studio. You need a systemic architecture: a cross-vertical handshake between your processes, your legal department, your metadata strategy, and your technical execution.

There are several different ways to build a compliance operation across a content supply chain, depending on the specific regulatory or commercial pressure you are facing. To illustrate how this architecture works, we will look at two examples built somewhat differently from each other. These examples demonstrate the scope of a modern operation and show how the handshakes between policy, process, and technology allow you to move through thousands of files a day instead of just a couple.

The Strategic Order of Operations: Revenue, Efficiency, and Shielding

Before you build the gates, you must define the strategic intent of the operation. A modern compliance operation is not a cost center; it is a strategic engine built with three specific goals in order of priority:

1. Revenue Preservation & The Engagement Multiplier: The primary goal is to enhance revenue. Advertisers and marketplaces are hyper-sensitive; unrated content is treated as high-risk, which prevents your ad-server from filling inventory. Beyond the ad-call, compliance is a growth strategy. Data from leaders like 3Play Media has proven repeatedly that captions increase watch time and comprehension. In a world where over 80% of mobile users watch on mute, missing captions is a direct attack on your viewership and your bottom line. If you keep revenue as your number one compliance goal, your teams will begin to see opportunity rather than roadblocks.

2. Shielding Against the Watchdogs: The second goal is mitigating risk against external threats. Accessibility non-profits act as industry watchdogs, actively monitoring streaming apps to find violations. They use litigation as a tool for progress, and they are the reason giants like Netflix and Hulu were forced to move ahead of the curve years ago. With the FCC’s $3.5 million fine against Pluto TV as a precedent, repeated lapses in closed captioning could be an extinction-level event for a small or medium-sized company.

3. The Liability Shift (The Corporate Shield): If the first two goals fail—if an asset slips through that is unrated or missing captions—your final goal is to shield the company. By using specific indemnity clauses in your acquisition contracts, you ensure the legal and financial fallout lands on the content source, not the distributor. In order to enforce a clause like that you not only need to have it stated clearly in the contract, but you also need to have a diligent and documented compliance operation that your legal team understand and are able to defend in a worst-case scenario.

Example 1: Standards and Practices (S&P) for Advertiser Safety

Many advertisers refuse to risk their brand by appearing next to TV-MA or Rated R content without explicit intent. However, the greatest threat to your revenue isn’t a mature rating; it’s an absent or incorrect rating.

The Rationale for the “Unknown”

In the programmatic ecosystem, ad-servers look for a rating to determine brand safety. If an asset is Unrated (NR), the server cannot guarantee it isn’t graphic violence or hate speech. Therefore, DSPs categorically exclude unrated content, rendering unrated effectively unmonetizable. Therefore, your system must treat unrated content as if it were the most sensitive rating possible (Above R or TV-MA). This ensures that your family-friendly ad tiers are protected by default.

The Holistic Lever System:

• The Contractual Policy: Your acquisition contracts should mandate that the Content Partner is responsible for delivering accurate ratings. This isn’t just a best effort clause; it should include indemnification language that shifts the legal and financial burden of a misrating back to the partner. This creates a powerful incentive for them to provide clean data at the source.

• The Human Process (O&O): For in-house content, there is no third party to blame. The process here requires producers, who are already intimately familiar with the material, to work with legal and revenue teams to apply a rating before the content ever hits the distribution supply chain. They are your built-in S&P team, removing the need for a separate department.

• The Data Role: Make sure you require a rating attached at the independent video asset level, not just at the series and episode level. The reason is that if, in the future, that content is repackaged, the content rating will be guaranteed to travel with wherever else that video is used. If the data is missing, the technology provides the final lever:

  1. Hard Fail: The ingestion errors out immediately.

  2. Soft Fail (Quarantine): The content is ingested but flagged. This flag ensures the content cannot be scheduled in the CMS. It remains “dark” and un-monetizable until the data is fixed.

Example 2: Closed Caption (CC) Compliance & The Data Conflict Method

Accessibility is your highest legal risk. Ensure captions are accurate, synchronized, and compliant with FCC and ACA standards.

The Step-by-Step Execution:

Step 1: The Contractual Policy (The Selects Phase)

Compliance starts during the acquisition negotiation. Require the partner to fill out a Catalog Sheet that serves as a legal disclosure:

1. Will captions be provided?

2. Are captions legally required by the FCC/CVAA?

   This allows your Finance and Operations teams to have the remediation conversation – who is paying for it and how long it will take – before the deal is even signed. While I strongly recommend to just require captions for 100% of your content, sometimes the real world doesn’t play accordingly, and this practice will help you avoid all kinds of issues from timing to cost to compliance.

Step 2: The Human Oversight Process

Your content operations team is the orchestrator here. They ensure that for O&O content, captions are sourced from high-quality vendors. They also oversee the communication loop with partners when the system detects a failure, ensuring that the business relationship remains intact while the technical standards are enforced.

Step 3: The Data Role (The Ingestion Logic)

When the media is delivered, your technology should compare the delivery metadata against the catalog phase metadata. You can choose two or three out of four specific data levers where you can choose a hard or soft error at ingestion:

1. Catalog Versus Delivery: If the catalog said “Required: Yes” but delivery says “Required: No,” the system flags a data conflict.

2. Required Versus Delivered: If the delivery said “Required: Yes” but “Included: No,” the system flags a data conflict.

3. The Presence Check: If metadata says “Included: Yes” but there is no file (e.g., .vtt) in the payload, the system flags a data conflict.

4. The Synchronization Gate: If the caption file duration does not match the video duration, the file should should be tagged with a soft error for manual review.

Step 4: Manual Review

During content processing, turn on the mezzanine file with the closed captions. Watch the first few seconds, the last few seconds, and about 1-3 other spots in the middle. This will enable you to check that the captions don’t drift, are the correct ones for that episode, and are of decent quality. In supply chains that I’ve designed this is done a the same time as spot checks for video and audio quality to save time.

Step 5: The Final Shield (The Reconciliation Report)

Your system should generate one definitive report for the partner per delivery. This is where the data meets the contract: “Please reconcile the following assets...and redeliver.” Don’t forget to cc the acquisitions team.

Conclusion: The Answer is the Operation, Not Just the Technology

Too many leaders in the streaming space attempt to solve compliance by jumping headfirst into technology solutions, by buying expensive AI monitoring tools or automated review software, hoping the software will do the work of a department.

The reality is that technology cannot accomplish this alone. Technology is a passive tool; only a cross-functional, orchestrated operation can achieve the goal of scaling a network safely. When the operation is humming in harmony – when your contracts dictate your data, and your data dictates your ingestion – there is actually very little technology required.

Building a compliance operation in the 100,000-asset era requires a combination of high-integrity data, policy-enforcing contracts, and a streamlined human process. You don’t need a massive S&P department; you need an Efficiency Machine that enhances revenue, mitigates risk, and shields the company from the ever-watchful eyes of the regulators and watchdogs.

Your transformation failed. Research from McKinsey and BCG has consistently found that around 70% of digital transformations fail to meet their objectives.

Here’s what may have happened: You chased technology instead of solving business problems. You tried to do everything at once. Your team was already stretched thin, and you asked them to execute a revolution on top of their day jobs. Leadership promised commitment and disappeared when things got hard. Or maybe it was none of those things and fortune just didn’t favor your ambition.

The good news? Failure is not fatal. It’s expensive, demoralizing, and embarrassing. It’s not fatal.

This article is for small to mid-sized streaming companies trying to figure out what to do next. Hard lessons from companies that recovered, the mistakes that killed the ones that didn’t, and a 90-day roadmap to stop the bleeding and start building momentum.

The path forward is not another sweeping transformation. It’s honest assessment, brutal prioritization, and small wins that compound. You don’t need to do everything. You need to fix one thing completely. Then do it again.

What Actually Went Wrong (And Why It Wasn’t the Technology)

The uncomfortable truth is this: it wasn’t the AI’s fault. It wasn’t the cloud migration’s fault. It wasn’t even AWS’s fault, though we’d all like to blame them.

Your transformation failed because you were solving the wrong problem. Or, more accurately, you were solving every problem at once.

The optics are alluring. Netflix generating billions from its recommendation engine. Disney’s hybrid monetization printing money. Competitors implementing AI everywhere. There’s real pressure to not fall behind.

So you greenlight a massive transformation. New infrastructure. AI personalization. Hybrid business models. Team restructuring. All at once. All urgent.

Here’s what the data says: Analysts estimate that roughly 70 to 84% of digital and business transformations fail to reach their stated goals. And for streaming companies specifically, you’re dealing with an industry that has spent years hemorrhaging money. Even Paramount, one of the largest legacy media players in the country, spent years fighting its way to streaming profitability. They reported 79 million global subscribers in Q1 2025, with streaming losses narrowing to $109 million that quarter, down from $286 million in Q4 2024. Full year 2024 streaming losses came in at approximately $497 million, down from $1.66 billion in 2023. After the Skydance merger closed and the operational restructuring took hold, Paramount+ hit roughly 80 million subscribers, and their D2C segment turned profitable in Q1 2026, generating $251 million in adjusted EBITDA. That’s a swing of $255 million from a $4 million loss in Q1 2025.

That took years. Real, grinding operational work. For a major studio with substantial resources. So if your transformation didn’t go the way you planned, you are in very good company.

The first failure pattern: technology-first thinking. You decided you needed an AI recommendation engine before confirming that content discovery was actually your problem. Maybe your users can’t find good shows because your catalog is shallow. Maybe your algorithm is fine. But you spent millions building an elegant and expensive solution to the wrong problem.

The second pattern: trying to do everything simultaneously. You attempted to upgrade infrastructure, implement AI, launch new business models, and restructure teams, all while your existing operations kept running, often in the middle of layoffs. You overloaded your best people. You created organizational chaos. And the data says that comprehensive transformations executed this way fail the overwhelming majority of the time.

The third pattern is the one nobody wants to talk about: culture. Studies consistently cite employee resistance and people-related factors as primary reasons why transformations fail. When half of your workforce feels unprepared for major technology-driven change and leadership charges ahead anyway, you get passive resistance, creative workarounds, and eventually, everyone abandons the new system. The tools get shelfed. The investment evaporates. And nobody says it out loud because they don’t want to be the one who killed the transformation.

One more thing worth naming: your technology stack probably looks like a patchwork quilt. Many streaming organizations run a double-digit number of tools and platforms across encoding, asset management, scheduling, advertising, analytics, and apps, creating a fragile, over-integrated stack. That’s a strategy problem that technology made worse.

The Reset: Clarity, Honesty, Alignment

Before you touch another product requirements document or schedule another transformation kickoff, you need three things. Not five. Three.

Clarity. What is the one business problem that is actually killing you? Not the ten problems you’d like to solve. The one that matters most.

Is it customer acquisition cost? Churn? Content discovery? Monetization? Pick one. Research on large transformation portfolios shows that a small minority of initiatives deliver a disproportionate share of total value, so focus matters far more than volume. The companies that recover are obsessive about their single biggest problem.

A mid-sized streaming company I know of spent about $3 million on a personalization AI that failed to reach its projected ROI. When they did the post-mortem, they realized their problem wasn’t recommendation quality. It was catalog depth. Users weren’t struggling to find good content in their library. There just wasn’t enough good content to match that particular audience profile. They solved the wrong problem expensively.

Honesty. Create a Transformation Graveyard. Document every failed initiative from the past three years. Study the patterns.

One team I know did this and found that all three of their major failures shared one thing in common: the executive championing each project left within six months of launch. Without sustained leadership commitment, transformations collapse. That’s it. It’s that simple.

This exercise is brutal. You are going to find things you don’t want to see. Insufficient budgets. Competing priorities. People who said yes in meetings and did nothing afterward. Do it anyway.

Alignment. You need someone whose actual job is driving this recovery. Not 20% of someone’s time. Not a committee. One person with authority, resources, and skin in the game.

For smaller companies, this might be a senior leader dedicating at least half their time. The research is clear: the people working on transformation need to be giving it at least half their time. Anything less, and you’re pretending.

The 90-Day Recovery Sprint

Forget the three-year roadmap. You need wins. Fast. Here’s how.

Days 1-30: Stop the Bleeding

Your first job is triage. What is actively broken? What is burning money? What is making things worse every day?

Stabilize it. This isn’t about fixing things properly. It’s about stopping active damage. Revert to old systems if you need to. Turn off features that are crashing. Pause initiatives that are creating chaos.

I’ve watched a team do exactly this after a codec migration went sideways. They had implemented a new delivery codec, and the versioning system was thrown into chaos by all the re-encoding. The entire content ID system became effectively useless, making it impossible to track revenue numbers or rights compliance. The first 30 days were simple: revert to the old system. Fix the pipeline. Get the site working again.

Was it glamorous? No. Was it necessary? Absolutely.

This is triage. You are stabilizing the patient so you can actually do the repair work.

Days 31-60: Pick One Battle and Win It

Now comes the actual fix. Choose one problem. Solve it completely. Not 80%. Not “good enough for now.” All the way through.

This is your proof that you can still execute. That you have learned from past mistakes. That you are capable of finishing what you start.

For streaming companies, high-impact wins often look like:

• Fix content ID integrity for revenue tracking and rights compliance on re-encodes. Test on all relevant versions across every platform. Consider applying a more flexible ID system upstream that’s more future-proof.

• Reduce buffering on mobile by 50%. Measurable, visible, completable.

• Improve ad targeting for one demographic segment. Clear scope, clear ownership, clear success metric.

When you’re solving a problem correctly, you need to think beyond the technology fix. The considerations that matter:

• Are your strategy and business objectives clearly defined and aligned with the fix?

• Are your operational workflows updated to support the new system, or are you rebuilding on top of the same broken processes?

• Are your metadata standards documented and enforced so the same problem doesn’t resurface?

• Does your team have the training and protocols to sustain what you’re building?

• Is your tech integration sound, meaning your delivery pipeline and encoding workflows actually respect what you’re putting in place?

Most failed transformations solved one part and ignored the others. That’s how you end up with a rebuilt content ID system that breaks again three months later because nobody trained the team on the new workflows.

Show your team and your board that you remember how to finish things.

Days 61-90: Build Momentum

With one win behind you, tackle a second initiative. Slightly larger. Still focused.

More important than the initiative itself: use this sprint to build the organizational muscle you’ll need long-term. Better project management. Clearer communication. Cross-functional collaboration that actually functions, not just a Slack channel where people ignore each other.

And do something that most teams resist: document your failures internally. Share what didn’t work. Build trust by being honest about the road you’ve been on.

What Comes After: The Long Game

After 90 days, you are not done. You are just beginning. But you should have stopped the bleeding, scored at least one real win, and started changing how your organization approaches technology initiatives.

Moving forward, three commitments matter.

Maintain discipline around scope. Every new initiative has to answer three questions before it gets a green light:

1. What specific business problem does this solve?

2. How will we measure success?

3. What are we NOT doing to make room for this?

If you can’t answer all three, you’re not ready.

Invest in your people. The streaming industry is racing to adopt AI. Fine. But technology without capability is expensive decoration. Budget for training. Create time for learning. Accept that building internal expertise is slower than outsourcing, but it’s the only thing that lasts.

Embrace iteration over revolution. The companies succeeding in streaming aren’t executing grand transformations. They’re making dozens of small improvements, learning constantly, and adjusting based on what actually works. The grand transformation is a great story to tell on stage at a conference. Iteration is what actually compounds.

The Real Difference in Streaming Transformations

The streaming industry is brutal. Failed transformations are common. They are not fatal.

The difference between companies that recover and companies that spiral isn’t budget size or technology sophistication. It’s three things:

Willingness to learn from failure. Discipline in execution. Recognition that transformation is a marathon of sprints, not a single heroic leap.

Start with 90 days. Fix one thing completely. Build trust. Score wins. Learn what works.

Then do it again.

The work is hard. The industry is unforgiving. Recovery is possible if you’re willing to stop pretending and start executing.

Your failed transformation taught you something expensive. Don’t waste the lesson.

This is a hard downturn. People have lost jobs. Projects have stopped midstream. Capital has pulled back.

It is the hardest operating period streaming has faced since the early digital transition, and it is showing who really understands their own systems. When the money slows, broken processes become more visible. The companies that fix them now will be the ones still standing when capital returns.

Research from Harvard Business Review shows that constraint drives innovation. Startups that take funding later tend to experiment longer and combine technologies in more original ways. Teams that work within defined outcome and time limits outperform those focused only on budget targets.

The same logic applies to streaming. Constraint is brutal, but it is not punishment. It is the natural governor that forces operators to get serious about how they run their business.

The Cost of Operational Drift

For years, streaming grew by expansion rather than control. More channels, more vendors, more dashboards. Every new partnership and integration added motion, but too few added structure.

That motion produced operational debt. Systems now overlap. Metadata pipelines are fragmented. Teams work around failures rather than repair them. Leaders manage activity, not alignment, which remains elusive at a lot of streaming networks.

Now the money has slowed, and the drift is visible. Redundant feeds erode margin. Manual reconciliation consumes time that should be spent on optimization. Basic workflows are still held together by Slack messages and institutional knowledge.

This is an operating model problem. The companies that will survive this period are those that treat integration as profit work.

What the Research Says

A 2025 Harvard Business Review study found that startups with large early funding rounds innovate less. Abundant cash leads to early scaling, reduced experimentation, and predictable technology stacks. Startups that received investment later kept experimenting longer and found novel combinations of existing tools.

MIT Sloan professors Fiona Murray and Elsbeth Johnson developed this further in their HBR piece “Innovation Starts with Defining the Right Constraints.” Their argument: budget and risk constraints narrow teams toward tried-and-tested solutions. Outcome constraints and timeframe constraints do the opposite. They force a fundamental reframing of the problem and produce genuinely new thinking. Teams given a clear target outcome and a real deadline consistently outperform teams given a budget ceiling and a list of risks to avoid.

The same dynamic plays out in media. When budgets are thin, teams have to ask sharper questions. What problem are we solving? Is this a structural fix or a short-term patch? Does this investment produce measurable value?

Constraint tightens focus. It forces every department to articulate purpose. That is where innovation begins, with clearer reasoning applied to well-defined problems.

What Constraint Looked Like at Pluto TV

Of all the startups I have worked in, only one achieved scale that lasted: Pluto TV. It succeeded because constraint was a core management philosophy, and it was constantly communicated from the top and throughout the company.

The rule was simple: spend money only where it increases company value. That meant intellectual property that created proprietary worth or user experience improvements that drove measurable engagement.

Internal tools were subject to a review process. Each request had to answer four questions:

1. Is this a short-, medium-, or long-term problem?

2. What is the cost-to-value ratio of solving it?

3. Does this add enterprise value or just internal convenience?

4. Can it be done with what we already have?

If the answer was unclear, the request stopped there.

At the time, that discipline felt frustrating. It slowed projects and required precision in justification. But it aligned the company. When resources are scarce, everyone learns to think as one system.

In practice, many of these requests were forwarded to me, where I would clarify the goal, the proposed process, the team, and the available tools. The goal was for me to outline how to solve the problem with what we already had available, rather than interrupt our roadmap. I would then send out the new process to the entire team and finish my email with this sentence:

“No engineering requests at this time.”

The discipline behind this policy helped Pluto grow efficiently, deliver a consistent user experience, and avoid the operational chaos that has since consumed other networks.

Why Streaming’s Dry Era Might Save the Industry

Scarcity is revealing what abundance concealed. For years, companies could hide complexity behind growth. Now, every inefficiency touches the balance sheet.

When budgets tighten, priorities sharpen.

• Redundant vendors get cut.

• Metadata systems are finally audited.

• Automation is applied where it belongs.

• Ownership becomes visible.

Constraint simplifies decision-making. It removes the illusion that more tools equal more capability. Every process that survives this era will be one that produces measurable value. That is what turning lean times into leverage means in practice.

No Cash, No Problem

Innovation does not start with a budget line. Some of the most creative operational solutions appear when teams are forced to work with what they already own.

If a workflow cannot be improved without new technology, it is often not understood deeply enough. If your metadata requires AI to reconcile, your metadata operation is the problem. Necessity builds skill faster than funding ever will.

This is the mindset shift the industry needs: Innovation only works well when it is a response to pressure. The companies that internalize that truth will reach profitability first.

A Practical Path Forward

Over the next six months, every streaming operator should focus on fixing the systems that are leaking money today. New initiatives will shape the future, but this is about protecting the foundation those initiatives depend on.

Fix strategic alignment. It’s free and it’s leaking money. Companies that fail to align strategy across departments lose roughly 30 percent of potential revenue before it ever reaches the profit line. The fix is communication, and it has to flow both ways along the org chart: top-down for direction and bottom-up for insight. Senior leaders set priorities, and every employee needs a path to surface what blocks progress. Skip-level meetings and executive memos are not a communication strategy. Real alignment happens when feedback loops are part of daily operations. When information moves freely in both directions, transparency and trust replace confusion. There is no software required for this, just leadership attention.

Define operational outcomes that move unit economics. Define what success looks like in operational terms. Tie every workflow and metric to corporate goals such as margin, engagement, or retention. If a project cannot connect directly to one of those outcomes, it’s just a distraction.

Set real deadlines and enforce them. Deadlines are real-world constraints with real-world consequences. Each operational fix should have a defined owner and a hard delivery date. “Ongoing initiative” is not a deliverable.

Collapse your identifier strategy. Choose one canonical ID system and map everything else to it. No exceptions. Every duplicate ID increases reconciliation cost and erodes data trust.

Freeze your taxonomy and document it. Every data field must have a single source of truth, a validation point, and a business owner. Taxonomies stabilize through ownership, not through committees.

Automate only what’s clean. Use AI only where pristine. Automation amplifies whatever exists. If the process is broken, it will multiply the error rate. AI does not clean data; it expands the mess. Automate only after the human process is stable, repeatable, and validated.

For Executives Who Know Their Operations Are Too Messy

If you cannot trace your content from ingest to playback, you do not control your business. If metadata does not reconcile across systems, your revenue reporting is unreliable. If automation requires human supervision, it is not automation. If six vendors sit between asset and audience, your costs will never fall, and you will never completely control your destiny.

These are structural and cultural problems. The longer they go unaddressed, the harder it becomes to return to profitability.

This era exposes what kind of operator you are. If you use it to document, simplify, and align, your company will gain speed. If you wait for new funding to hide the mess again, it’s not coming anytime soon.

Downturns Don’t Decide Who Survives. Operations Do.

The dry era is painful, and it deserves honesty. People are carrying heavier loads with smaller teams. Projects are under scrutiny. Every investment needs proof.

But inside that pressure is a rare opportunity. You have permission to rebuild the system the right way. To clean the data, simplify the process, and demand accountability for how value moves through your company.

Money will return. When it does, it will amplify whatever system you have in place. If that system is clean and connected, the money becomes margin. If it is still chaotic, it becomes another round of waste.

Streaming’s dry era sucks. Use it to build operations that don’t.

The biggest revenue leaks in media companies don’t come from technology failures or metadata gaps. They come from misaligned goals, unfocused meetings, and communication habits that fragment attention. These are also the easiest leaks to fix, and they don’t cost a dime. By resetting meeting culture, aligning next-year goals, structuring communication, and auditing project ownership, CEOs can contain real money this quarter.

Why Zero-Cost Levers Matter

Revenue leaks in media companies come from many sources. Some are technical, like metadata mismanagement, inefficient workflows, sprawling vendor stacks. But what every major consultancy and research firm agrees on is this: the largest leaks are not operational or technical. Lack of aligned strategy and internal communications alone can cost companies 30% of their earned revenue before it hits the profit line.

Projects launch without a clear line to enterprise goals. Teams chase activity that looks valuable but doesn’t tie back to growth. Communication channels reward noise over clarity, and the most important work goes underpowered.

This is where CEOs have unique leverage. The authority to reset expectations, the influence to change cultural signals, and the discipline to hold leaders accountable all sit at the top. Unlike capital investments or reorganizations, these moves cost nothing.

Here are four levers I have seen CEOs pull, with immediate impact.

1. Put Meetings Back in Service of Strategy

Meetings are the stealth tax on every company. They look free, but they carry a real cost: hours of salaried attention pulled away from value creation. A weekly executive meeting of ten leaders costs six figures annually in payroll. Multiply that across the organization, and the drain becomes undeniable.

When I was at Pluto TV, we instituted a cultural rule that still stands out as one of the most effective I have ever seen: if you are in a meeting where you are not contributing or learning, you are expected to leave.

The premise was simple. Staying in the room when you were not adding value was wasteful. Time in a meeting is company money. Walking out was not rude. Staying was.

The results were immediate. Meeting rooms emptied of passive bodies. Conversations sharpened, because the only people left were those directly engaged with the topic. Leaders reclaimed hours each week for actual execution. And the cultural tone of what a streaming company should be was upheld. Time was treated as a scarce, valuable asset.

If you want to set this standard in your own company, start with three moves:

Require every meeting to have a goal or agenda upfront.

Normalize walking out when you are not learning or contributing.

Collect feedback on recurring meetings. Do they serve strategy, or are they just slideshows that nobody is paying attention to?

The CEO’s visible endorsement matters here. When your team sees you enforce these rules, meaning when they see you leave meetings yourself, they know the standard is real.

2. Align Next-Year Goals Before Resources Are Spent

The second major source of leakage is misaligned ambition. By the time departmental budgets hit your desk, most of the work has already been scoped. Waste feels baked in. The smarter move is to surface goals earlier.

Ask each vertical leader now for their next-year goals in writing. Then hold them up against your company’s strategic KPIs. Where alignment is strong, give backing. Where goals drift into pet projects, duplicative efforts, or initiatives with no clear line to revenue, require a reset before resources are committed.

This step does two things:

It prevents funding work that dilutes focus.

It forces leaders to prove how their ambitions support enterprise growth.

The alignment conversation can be uncomfortable, but it pays immediate dividends. You discover where leaders are already pulling in the same direction and where they are not. You also prevent months of investment in projects that were never going to generate returns.

Simple alignment at the front end avoids millions wasted on phantom ROI.

3. Structure Communication for Deep Work

If meetings are a tax, communication is often an invisible siphon. Slack, Teams, and email create the appearance of alignment but often generate duplication, distraction, and shallow work. Busy, but not productive.

The real leak is not the tools. It is the unstructured way they dominate attention. When your most talented people spend their peak energy answering redundant questions in Slack, they are not building products, closing deals, or solving problems.

Three structural shifts can change this:

Centralize documentation. Create a clear, accessible hub for workflows, processes, and decisions. Every documented answer prevents a cycle of repeat questions.

Audit communication load. Look honestly at how much messaging could be solved by documentation. Look at how often decisions stall because information is scattered.

Protect deep work. Encourage teams to block time when Slack or email is shut off. Make it clear that productivity is measured in outcomes, not responsiveness.

Leadership tone matters here. When the CEO says, “Time spent on deep work is time well spent,” behavior shifts.

I have seen entire product roadmaps regain momentum once teams were freed from the constant pressure of digital chatter. The cultural permission to ignore the noise channels talent into creating outcomes instead of notifications.

4. Audit Ownership of Major Projects

Projects without clear ownership are one of the most expensive forms of waste. They drag on, consume budget, and often deliver little because the accountability structure was never defined.

An ownership audit is one of the simplest, most effective tools a CEO can use. For every significant initiative, ask three questions:

Who is the business owner?

What is the single-sentence value statement for this project?

Is ownership aligned with the people who will actually use the outcome to achieve their goals?

If any answer is unclear, pause the project until ownership is reset.

I have seen multi-million-dollar initiatives reverse course once ownership was clarified. A single question, “Who actually owns this?” has the power to stop waste in its tracks.

The pattern is consistent across organizations. When projects lack clear ownership, they metastasize. When ownership is defined and aligned, they deliver.

Leadership Attention Is the Cost

Revenue containment is not about austerity. It is about protecting enterprise value from silent erosion. The most powerful moves available to a CEO cost time and attention, not capital.

By reclaiming meetings, clarifying goals, structuring communication, and auditing project ownership, you establish a culture where work serves strategy, communication advances execution, and projects deliver outcomes.

I have seen companies save millions by adopting these practices. They did not get there by spending more. They got there by insisting on clarity.

Streaming companies are leaking revenue not because they lack strategy, but because the strategy never reaches the people who execute it. McKinsey research puts roughly 30% of organizational performance at risk from strategic misalignment. PwC estimates coordination breakdowns cost companies up to 350 hours per employee per year. The result is invisible on a financial statement, but the damage is real: missed launches, redundant work, slow decisions, and teams that eventually stop trying to figure out what the company actually wants. Misalignment compounds across quarters until growth stalls and everyone blames the market.

The Hidden Cost of Strategic Drift

In streaming, strategy moves fast. Markets shift, partners change, distribution windows open and close in weeks. But the people doing the work, the producers, ops leads, engineers, marketers, and analysts, hear the strategy late or not at all. By the time it filters through three layers of leadership, the original intent has been rewritten, softened, or abandoned.

This is what strategic misalignment looks like in practice. Plans exist; alignment doesn’t. And in streaming, where speed and coordination drive margins, that gap is a revenue killer.

I’ve seen it most sharply in organizations going through rapid scaling or post-merger integration, the kinds of environments where headcount doubles before the org chart catches up. The strategy exists somewhere at the top. What lives in the operating layer is a collection of interpretations that have drifted far enough from the source material that teams are essentially working from different documents.

How Top-Down Communication Breaks

Most streaming organizations run a cascading model: executives announce the strategy, VPs interpret it, directors translate it, and managers operationalize it. Each layer adds context, filters detail, and introduces drift.

By the time the strategy reaches the people executing it, three things have usually happened:

1. The original “why” has been replaced with a “what” or a “how.”

2. The strategic priorities have been re-ranked to match each leader’s domain bias.

3. Cross-functional dependencies have been flattened into siloed deliverables.

What started as a clear directive becomes a series of disconnected initiatives. Teams optimize for their slice of the problem and miss the system-level goal.

I watched this play out during a content expansion push at a streamer that was moving fast to fill genre gaps. The executive directive was clear at the top: accelerate premium sports-adjacent programming to drive subscriber retention in key demos. By the time it reached the production operations team, it had become “clear the bottlenecks in unscripted.” The teams were busy. They were not building the thing leadership thought they were building.

Cross-Functional Misalignment in Streaming

The streaming business is a particularly brutal test of alignment because no single team can ship value alone. Programming, ad sales, product, engineering, and operations all touch the same content, the same audience, and the same revenue stream. When they’re misaligned, the consequences compound.

Here’s what that looks like in practice, from environments I’ve worked in or alongside:

• Programming launches a new genre block. Ad sales hasn’t priced inventory for it.

• Product ships a personalization feature. Editorial doesn’t have the metadata to populate it.

• Operations consolidates a workflow. Finance still tracks it as two cost centers.

• Marketing promotes a tentpole title. Engineering’s CDN can’t handle the traffic.

• Acquisitions locks a new window deal on library content. Streaming ops finds out when the content doesn’t show up in the CMS on launch day.

Each of these reads like an execution failure. It isn’t. These are strategic failures wearing execution clothes. The teams are doing exactly what they were told. They’re working from different versions of the plan.

At scale, the compounding gets ugly. A FAST channel launch that should take six weeks takes fourteen because five teams are sequencing correctly within their own systems while nobody is running the cross-functional clock. The delay isn’t anyone’s fault in isolation. The fault lives in the absence of a shared operating picture.

Town Halls Are Not Communication

Most executives believe they’re communicating strategy because they hold quarterly town halls and circulate a deck. A town hall is a broadcast. It’s a performance moment, not an alignment system.

Real strategic communication is continuous, multi-directional, and verifiable. In practice that means:

• Executives test understanding rather than just transmitting information.

• Middle managers translate strategy into team-level decisions, with enough specificity that people know which tradeoffs to make.

• Individual contributors can articulate how their work connects to the company’s top three priorities.

• Feedback loops surface drift before it becomes damage.

The test I use: walk into any team meeting unannounced and ask someone to describe what the company is trying to accomplish this quarter and why their work connects to it. If the answer is vague, misses the “why,” or reflects a priority the executive team deprioritized two months ago, there’s no alignment; there’s a deck.

I’ve sat in operating reviews at companies where the slides in the room and the strategy on the CEO’s website were describing two different businesses. Nobody had updated the field. The field was still executing against the old thesis. Months of work, significant headcount, real dollars, all pointed at an objective the leadership team had already moved on from.

What Misalignment Costs

The numbers are easy to dismiss because they don’t appear as a line item. They’re real:

• 30% of organizational performance lost to misalignment (McKinsey).

• 350 hours per employee per year wasted on coordination friction (PwC).

• Launches delayed by weeks because dependencies weren’t surfaced until the week before go-live.

• Margin erosion as redundant work and rework accumulate across teams.

• Talent attrition as the best operators burn out carrying a system that should be running itself.

That last one matters more in streaming than in most industries. Streaming ops talent is specialized and scarce. When a strong director-level operator leaves because she’s spent eighteen months trying to align teams that leadership won’t invest in aligning, the cost shows up in recruiting fees, ramp time, and institutional knowledge walking out the door. None of it appears in the alignment budget because there isn’t one.

Alignment shows up directly in the P&L. In streaming, where the operating model is the product, misalignment is one of the most expensive problems a company can carry.

What Good Looks Like

The streaming organizations I’ve seen run well share a few habits that are less about management philosophy and more about operational discipline:

1. Narrative compression. The strategy fits on a single page. Anyone in the company can repeat it without needing the deck.

2. Cross-functional dependency maps. Every team knows what it owes other teams and what other teams owe it, before the quarter starts, not during it.

3. Explicit decision rights. Who decides, who is consulted, and who is informed is written down. Assumed decision rights get contested at exactly the worst moment.

4. KPIs that match strategy. What gets measured matches what the executive team actually wants to happen.

5. Continuous calibration. Leaders test understanding monthly, not quarterly. A quarterly check-in finds drift after it’s already expensive.

These are operational hygiene. The companies doing them aren’t smarter than the ones that aren’t. They’re more disciplined, and they’ve usually been burned before.

Pluto TV in its early scaling years was relentless about cross-functional rhythm. The business was moving too fast to afford misalignment, so the leadership team built lightweight but consistent mechanisms to keep teams pointed at the same thing. That discipline was part of what made the growth defensible.

The Bottom Line

Strategy is only as strong as the clarity with which it travels through an organization. Knowing that the company hit its revenue target doesn’t explain which moves worked, and most streaming organizations are operating in exactly that fog: hitting numbers without a clear read on causality, and missing numbers without a clear read on what broke.

The fix isn’t a consultant or a culture initiative. It’s leadership deciding to treat communication as a system rather than a calendar event, and then holding the operating structure accountable for alignment the way it holds teams accountable for delivery.

Misalignment is expensive, pervasive, and solvable. The companies that solve it aren’t waiting for a better strategy. They’re making the strategy they have actually work.

Most media teams believe their metadata is good enough. Assets are findable. Rights are mostly tracked. Delivery issues usually get caught before air.

I’ve been inside enough operations to tell you what “mostly” and “usually” are actually costing you. Every gap in your metadata is leaking money, consistently and across multiple departments. It shows up as missed ad revenue, delayed platform launches, partner friction, and cycles of rework that are more expensive than anyone wants to admit.

These are not minor inefficiencies. They are structural drains. And most of them are fixable without rebuilding your CMS from scratch or launching an eighteen-month taxonomy overhaul.

Here are five tactical changes you can make to how you think about and use your content metadata, with immediate impact.

1. Map Strategy to Execution with Durable Metadata

Your content strategy should not live only in a slide deck or someone’s head, but that is exactly where it ends up when metadata does not carry the signal.

When metadata reflects only short-term or reactive priorities, like “this week’s push” or “today’s urgency,” teams end up buried in disposable tags that expire faster than they can be used. The strategy lives at the executive level and never makes it to execution. The two layers keep operating out of sync, and no one can figure out why alignment is so hard.

**The fix:** Create durable metadata fields that anchor to long-range business objectives. Franchise flags, monetization categories, revenue tiers, audience priorities. These are the connective tissue that carries your strategy from kickoff through delivery. Temporary programming notes belong in a clearly separate field or annotation layer, not tangled up with the structural data your teams depend on.

Think of metadata as the connective signal between executive strategy and operational action. If it only points to this week’s priorities, your teams are flying on dead reckoning. That gets expensive fast.

2. Clean Up Your Avails and Publish Data

Advertisers are not simply buying blank inventory. They are buying context. And they are increasingly reliant on descriptive metadata to make targeting decisions that actually hold up.

If your avails metadata is inconsistent across platforms, or if you are skipping qualitative descriptors like tone, topic, and sentiment, you are leaving money on the table. The mechanism is not mysterious: programmatic systems need metadata signals to match ads to inventory. When those signals are missing, the match does not happen. The inventory goes unfilled or undervalued.

The data on this is concrete. Gracenote’s Global Video Data analysis of FAST programming found that only 35.4% of sports programming headed to FAST channels in the U.S., Great Britain, Germany, and Canada included original air date information when submitted for enrichment. That missing signal makes it impossible for programmatic systems to identify live sports competitions as premium inventory, which is exactly what live sports is. Without the tag, the system cannot see what it is looking at. And when the system cannot see it, you cannot sell it at the right price.

This is a broader data quality problem, not just a media one. Gartner estimates that poor data quality costs the average enterprise $12.9 to $15 million annually, with 20 to 30 percent of enterprise revenue lost to data inefficiencies. In media, those inefficiencies are hiding in the metadata your AdOps team is trying to trust.

Here’s how to fix it: Standardize your avails metadata, and enrich it with high-quality descriptors that make your content easier to match, package, and monetize. What used to be optional enrichment data is now the baseline expectation for programmatic buying. If your AdOps team cannot trust the feed, your ad inventory will consistently underperform.

3. Make Rights Metadata Work for You

Rights errors are metadata problems. They are almost always at least partly metadata problems. Launches get delayed. Takedowns happen mid-campaign. Entire content categories go underutilized because no one is confident about what is cleared for what.

I wrote a longer piece on this topic, but it bears repeating. Your expensive RMS has to have accurate, complete data to work with. If the rights metadata is incomplete or siloed from the rest of the content workflow, no platform is going to fix that for you.

What to do instead:Embed rights metadata directly into your content workflow. Platform, territory, format, start and end dates, holdbacks. Bake it in early, and enforce it through automation wherever possible. Memory, guesswork, and Slack threads do not constitute a rights system. They constitute a liability, and a slow-moving one at that.

If your rights metadata is not part of the content’s DNA from day one, you are counting on people to remember things that should be structured fields. That works fine until it doesn’t. And when it doesn’t, it is always at the worst possible moment.

4. Track Status and Source to Avoid Rework and Delivery Delays

Delivery failures are rarely a technology problem. They are a communication problem. Teams do not know which version is final. They do not know what work has already been done. They are not sure which asset is headed to which partner. So they guess. And then they redo.

The rework is expensive. The chaos is avoidable.

The approach that works: Use metadata to track asset status, format compliance, and point of origin. Much of this can and should be automated. When you automate source and status tagging, you remove the guesswork from the delivery process and give every team member a single, reliable source of truth. Without that tagging, your delivery workflow becomes a high-stakes game of telephone. The cleanup is expensive, and it compounds over time.

This is one of the places where a relatively small investment in metadata infrastructure pays back quickly. You do not need a new system. You need the system you have to carry better signals.

5. Connect Metadata to Money

When metadata stops at the editorial layer, you are flying blind on cost. Content that requires extra editing, reformatting, clearance work, or localization often absorbs overages that never get reconciled against the revenue it generates.

Multiply that across hundreds of assets per quarter, and you have an invisible margin problem. The operations team absorbs costs that should be visible. Leadership makes decisions about content investment based on incomplete financial data. And the pattern repeats.

To close that loop: Introduce fields that capture overages, unexpected prep work, and exceptions to the standard pipeline. Tag assets with projected revenue stream and licensing or payment structure where applicable. Over time, this builds a clear financial picture of asset-level ROI, which is something most content operations teams cannot produce today.

When you do not tag the overages, you absorb them. And that turns your operations function into a financial black box that leadership will eventually stop trusting, if they have not already.

Where to Start

If your content business is leaking time, money, or operational clarity, start by looking at your metadata. If you are considering a technical overhaul, know that you need to harness your metadata first to avoid a failed transformation.

These five areas do not require a massive overhaul to address. They require intentionality, a clearer field structure, and a willingness to treat metadata as infrastructure rather than administrative overhead. I have seen each of these fixes reduce rework, accelerate delivery, and recover revenue that was already being earned but not fully realized.

Metadata is not glamorous. It runs the show behind every deal, delivery, and dollar anyway. Treat it like the asset it is, and your bottom line will reflect it.

Prologue: Why I’m Telling You This

2026 marks twelve years that I’ve been doing the work of recovering from and managing PTSD. Four years of trauma therapy, executive coaching that followed, and a master’s degree in organizational management and leadership: all of that brought me to a particular kind of clarity about how human nervous systems behave under sustained pressure.

I entered and rose in the corporate world carrying that history. The more advanced I became at cross-functional leadership, the more I noticed something I couldn’t unsee: the same trauma patterns I had spent years learning to recognize in myself were playing out at scale inside companies. Manipulation masking as management. Arbitrary reorgs destabilizing high-performing teams. Thick silence in executive meetings underscoring tensions no one would name for fear of retaliation.

What I’m sharing here is the operational version of something I learned in my own recovery. Trauma doesn’t only live in people. It embeds in systems. And systems, like people, remember.

Years of Mass Layoffs

The scale of layoffs across tech and media over the past several years has been genuinely historic, and it has not stopped.

In tech, 2022 marked the beginning of a massive correction. Roughly 165,000 tech employees lost their jobs that year, as companies reversed the aggressive pandemic-era hiring that had ballooned their headcount far beyond sustainable levels. The downsizing intensified sharply in 2023: approximately 262,700 tech workers were laid off, a surge of 59% over the prior year. By some tallies, over 460,000 tech employees globally had been cut since the start of 2022, an astonishing reversal for an industry that, just months earlier, had been competing desperately for the same talent it was now discarding.

2024 saw some easing. 2025 did not. AI-related layoffs alone accounted for roughly 55,000 cuts that year, and November 2025 recorded more than 71,000 layoffs across the broader job market, the second-highest monthly figure in five years, according to Challenger, Gray and Christmas. Then 2026 arrived and the numbers surged again. By April 2026, tech layoffs had already surpassed the full-year pace of 2025. Meta announced a 10% workforce reduction. The cuts kept coming.

Media followed its own version of the same arc. In 2023, U.S. media companies announced more than 21,000 job cuts, the highest annual total since the 2008-09 recession (excluding the extraordinary disruption of 2020). For context, fewer than 4,000 media jobs had been cut in all of 2022. The acceleration was brutal. By mid-2023, media layoffs had already exceeded even the worst months of the early pandemic. Newsrooms large and small were gutted. BuzzFeed shuttered its entire news division. CNN, the *Los Angeles Times*, and dozens of others slashed substantial portions of their staffs.

2024 brought a slight dip. 2025 did not hold that line either: more than 17,000 entertainment and media positions were eliminated in the first eleven months of 2025 alone an 18% increase from 2024’s already punishing numbers. Then, in March 2026, CBS News cut 6% of its workforce. WME cut 3% the same week. Lionsgate, Axios, Universal Music Group, and Netflix’s global product team all announced cuts within days of each other. Consolidation was cited as the driver each time: the Paramount/Skydance merger, the Warner Bros. Discovery acquisition of Paramount, the relentless logic of cost reduction in a streaming ecosystem that over-expanded and is now contracting.

The industry has been doing this on a loop. And the loop has consequences that the industry has largely chosen not to name.

The Human Toll

Losing a job is not a career setback in the way a missed promotion is a setback. The research is unambiguous: layoffs are one of the most stressful life events a person can experience. One study ranked being laid off seventh on the list of life’s most stressful events, placing it above divorce and above the death of a close friend. Mental health experts put the average psychological recovery time at roughly two years. That is a long time to be unstable.

The physical health consequences are equally serious. For otherwise healthy individuals, involuntary job loss increases the odds of developing a new serious health condition by 83% in the first 15 to 18 months after the layoff, according to research cited by Chelsea Green Publishing. Stress-related conditions spike: hypertension, cardiovascular disease, autoimmune disorders. The psychological and financial pressure carries even graver risks. Stanford’s Jeffrey Pfeffer has documented that layoffs increase the odds of suicide by two and a half times, and increase mortality by 15 to 20% over the following 20 years. “Layoffs kill people, literally,” is how he put it.

Displaced workers are twice as likely to fall into depression, four times more likely to abuse substances, and six times more likely to commit violent acts, including domestic abuse, according to findings cited by the *Harvard Business Review*. The U.S. Department of Labor has explicitly acknowledged that being laid off is one of the most traumatic experiences a person can go through. The damage doesn’t stay contained to the individual. It radiates outward into families, households, and communities, creating public health costs that never appear on any company’s balance sheet.

The multi-round dimension makes all of this worse. When companies conduct a second round of cuts after the first, the negative impact on surviving employee reviews is roughly twice that of a single-round layoff, according to analysis by Onwards HR. The people most affected are the ones companies can least afford to lose: key talent, managers, and recent hires who haven’t yet had time to develop the institutional cynicism that inoculates some long-tenured employees against caring too much.

What Happens to the People Who Stay

I’ve been on both sides of this. I’ve been laid off. I’ve also been the person who survived rounds in organizations going through them, which carries its own particular weight that rarely gets named honestly.

The survivors of a layoff are not the people who escaped harm. They are the people who watched colleagues and friends get walked out, absorbed the remaining work, and are now expected to generate forward momentum for a company that just demonstrated it views headcount as a cost lever. The gap between what leadership communicates and what the remaining team actually experiences is usually enormous. The company calls it a reset. The people in the building call it what it is.

A Harvard Business Review study of 146 companies that conducted layoffs between March 2020 and November 2022 found sharp, measurable declines across three core dimensions of employee engagement: company confidence dropped 16.9 percentage points, belief in career opportunities dropped 12.1 percentage points, and confidence in leadership dropped 10.5 percentage points. HBR research also found that engagement typically takes 12 to 24 months to return to pre-layoff levels, assuming nothing else goes wrong in the interim.

The productivity numbers are worse than most companies want to hear. A Leadership IQ survey of more than 4,000 employees who had survived a layoff found that 74% said their own productivity had declined, and 87% said they were less likely to recommend their organization as a place to work. That last number is a recruiting metric and a brand metric. It has a cost. Glassdoor ratings drop 0.13 stars after a layoff, and the data shows it takes approximately 32 months for those ratings to recover. A 2024 Gartner survey added to this picture: 57% of employees said morale fell after layoffs, and 53% said their trust in leadership eroded.

Survivor’s guilt is real, and it goes underestimated. So is the fear that runs through an organization after the first round: a specific, hypervigilant preoccupation with whether another round is coming, when it might arrive, and who will be on the list. A 2023 industry survey of tech employees found that 77% reported their mental well-being had deteriorated from the sustained fear of layoffs. That is not a morale problem. That is an organizational health problem.

When an employer cuts a substantial portion of its workforce, it breaks something specific and measurable: the psychological contract. That contract is the unspoken agreement between employer and employee that says, broadly, if you perform and contribute, you will have security and a future here. When it breaks, the remaining workforce recalibrates its relationship to the company. People stop volunteering discretionary effort. They protect themselves. They start looking.

What Happens to the Company

Mass layoffs may be executed in the name of efficiency, but they routinely hollow out the organizational capabilities that efficiency depends on.

Harvard Business School professor Sandra Sucher, who has studied the hidden costs of layoffs for years, documents what happens to the people who leave: they take with them years, sometimes decades, of institutional knowledge, customer relationships, and operational context. That knowledge doesn’t transfer in an offboarding call. It walks out and often ends up benefiting a competitor. The organization is left thinner in ways that don’t show up on the balance sheet until the next critical project hits a wall that the remaining team doesn’t know how to get around.

Innovation suffers in measurable ways. After a layoff, the remaining staff tends to become more risk-averse. The implicit message has been received: this company will cut when it needs to, which means advocating for a long-shot idea or a bold initiative is a riskier personal proposition than it was before. People who are uncertain about their own continued employment do not take risks. They protect their standing. Sucher notes that overall innovation and new product development reliably suffer in the wake of staff cuts, and that the recovery, if it comes, can take years.

Pfeffer’s research complicates the financial logic executives cite when announcing cuts. Workforce reductions frequently do not reduce costs in practice: severance packages are expensive, unemployment insurance rates climb, and companies often end up rehiring laid-off workers as contractors, paying the contracting firm a premium for people they already trained. Any immediate payroll savings are frequently offset within three years by downstream costs including higher voluntary turnover, declining morale, and customer-facing degradation of service or product quality. The expected cost savings are often illusory. The organizational drag is not.

The trust damage is structural. A culture of silence takes hold after a layoff. People stop raising candid concerns. They stop proposing ideas that might make them visible in ways they can’t predict. Leadership can simultaneously grow more disconnected, retreating from inspiration into operational brass tacks, which accelerates the engagement problem rather than containing it.

Why Layoffs Got Normalized Anyway

Given all of this evidence, the question worth asking is why mass layoffs became the default response to financial pressure across two of the most sophisticated industries in the country.

The first answer is Wall Street. A layoff announcement signals to financial markets that management is taking action on costs. That signal, at least in the short term, often produces a stock price bump. This dynamic has been documented for decades. *Newsweek* noted as far back as the 1990s that “the more people a company fires, the more Wall Street loves it, and the higher its stock price goes.” Once that incentive calculus became clear, layoffs shifted from a last resort to a routine earnings management tool. CFOs began calculating immediate payroll savings without accounting for the organizational drag and talent drain that would follow. Traditional accounting practices don’t capture the cost of lost morale or lost institutional knowledge, so those factors get excluded from ROI calculations.

The second answer is what Pfeffer calls social contagion. When high-profile firms announce cuts, others follow, often not because their own situation requires it, but because their peers are doing it and doing nothing looks imprudent. During 2022-2023, as Meta, Google, and Disney announced major layoffs, a herd mentality settled into Silicon Valley and the media sector. Layoffs became a badge of fiscal discipline even when companies remained profitable. Pfeffer documented this directly: “Meta has plenty of money. These companies are all making money. They are doing it because other companies are doing it.” Copycat behavior at that scale inflicts damage that is neither necessary nor traceable back to any genuine strategic requirement.

The third answer is language. Calling a layoff a “reduction in force,” a “right-sizing,” or an “organizational optimization” puts distance between the decision and its human consequences. That distance makes the decision easier to make repeatedly. It also makes it easier to avoid the harder work of diagnosing whether the underlying problem is strategy, market positioning, overexpansion, or something else entirely. Pfeffer has argued that layoffs are often “a confession of poor management,” an admission that leadership failed to plan ahead, overbuilt during growth, and is now reaching for the bluntest instrument available to correct it. Layoffs belong in the category of last-resort interventions, not routine cost management. The normalization of the practice is, itself, a form of institutional failure.

Top-Line Leaders Are Not Okay Either

Let’s not pretend this only happens at the team level.

Trauma climbs.

In the past several years, C-suites across media and tech have absorbed more organizational shock than they know how to metabolize: economic upheaval, mass layoffs, failed pivots, shifting mandates, relentless pressure from boards, markets, and media coverage of their own decisions. Many leaders, rather than processing that pressure, have begun projecting it.

The pattern is visible. The shift in tone. The resurgence of hard-nosed, top-down management. Executives declaring that “the era of nice is over.” RTO mandates rolled out in waves. Flexibility revoked. Control tightened at precisely the moment when the workforce most needed some degree of autonomy to feel stable. The instinct is recognizable to anyone who has done trauma work: when you can’t tolerate the discomfort of uncertainty, you reach for control. It is faster than sitting with the discomfort. It is also less effective.

I don’t think most of these leaders are malicious. I think they’re scared. I think they’re trying to normalize the damage, frame it as discipline, and call it culture. The pattern mirrors how many people cope with their own unresolved stress: by compartmentalizing it, rationalizing it, or rebranding it as rigor. Unprocessed pressure at the top has ripple effects throughout the entire structure below it.

Fear at the top creates fear everywhere else. The nervous systems of organizations are always downstream from leadership. If leaders are locked in self-protection mode, so is everyone else. Creativity flattens. Collaboration crumbles. Systems grind.

Leaders who are willing to name the harm, own their part in it, and lead from clarity rather than from defensive hardening create the conditions for real recovery, for themselves and for their organizations. That is the harder path. It is also the more effective one.

What Recovery Actually Requires

More voices in business have begun calling for a rethinking of how mass layoffs are used, and that is worth acknowledging. But the frameworks being offered are often too thin for what is actually needed.

Some companies have begun exploring alternatives to headcount reduction when financial pressure arrives: reduced hours, temporary pay cuts across the organization rather than permanent job cuts concentrated in specific roles, hiring freezes, voluntary attrition programs, or re-skilling investments that redirect existing employees rather than eliminating them. Buffer and other organizations have gone public with decisions to hold hours and cut executive pay rather than reduce headcount. Lincoln Electric famously held its workforce through downturns by distributing reduced hours broadly rather than concentrating the impact on a smaller group. Southwest Airlines declined to lay off after September 11, 2001, and gained market share by the end of that year. These are not fringe positions. They are operational choices with documented outcomes.

For organizations that have already gone through layoffs, the recovery work is slower and more specific than most post-layoff communication plans suggest. You cannot fix a broken psychological contract with a town hall. You cannot address the trust collapse with a CEO note about resilience. What the research consistently shows is that the companies that recover fastest are not the ones that move on quickly. They are the ones that slow down long enough to be honest about what happened, that don’t force people into a galvanizing narrative before the grief has moved through, and that demonstrate over months of consistent behavior that leadership has actually learned something.

That means transparency as a default, not a communications strategy. It means involving surviving employees in planning the path forward rather than delivering it to them as a finished document. It means naming what failed, plainly, without the protective varnish of HR language. It means investing in the development of the people who stayed rather than treating them as a retained cost on a compressed budget. And it means understanding that the 32-month Glassdoor recovery timeline and the 12-to-24-month engagement recovery timeline are not anomalies. They are the price of the decision, and they need to be accounted for.

Gartner research is direct about the mechanism: survivors re-engage when leadership uses actions over words. Not communications. Actions. Consistently, over time. There is no shortcut for that part.

Why I Keep Writing About This

This piece started, years ago, as an attempt to name something I kept watching happen inside organizations and couldn’t find anyone else writing about with any operational precision. The trauma literature existed. The layoff research existed. But the connection between what happens inside human nervous systems under sustained threat and what happens inside company cultures under sustained layoff pressure: that was a gap. It still mostly is.

I write about this because I’ve been on multiple sides of these decisions. I’ve survived layoffs. I’ve sat close enough to C-suite decisions to understand the pressures that produce them. I’ve watched organizations flatten, fragment, and lose years of built capability in the time it takes to complete a single restructuring. And I’ve watched leaders struggle with the aftermath in ways they weren’t given tools to understand because the dominant framing of layoffs as a financial decision strips out most of what matters about them as an organizational event.

The through-line from my own recovery work to this writing is a simple one: when you can’t name what’s happening to you, you can’t respond to it effectively. Organizations that can’t name what a layoff actually does to the people and systems inside them keep making the same decisions, absorbing the same costs, and wondering why the recovery takes so long. Naming it clearly is the first step. Everything else follows from that.

AI scales whatever operational reality it’s given. Hand it a well-governed, cleanly instrumented pipeline, and it multiplies your output. Hand it a mess of inconsistent metadata, undocumented workflows, and informal ownership, and it multiplies that too. The companies that are winning with AI in media operations didn’t get lucky. They did the foundational work before they plugged anything in.

AI Scales What You Already Are

AI doesn’t fix broken processes. It automates them. That’s the part most deployment conversations skip.

When a media organization decides to “adopt AI,” the conversation almost immediately jumps to which tools to evaluate, which vendors to pilot, which use cases to prioritize. Rarely does it start with: what are we actually feeding this thing? What does our metadata look like? Who owns it? How was it generated? What’s the error rate?

Those questions are unglamorous. They don’t make for good conference slides. But the answer to every single one of them will determine whether AI becomes leverage or liability.

The real question is whether your infrastructure is ready to carry more weight. If your existing workflows are undocumented, your metadata is inconsistent, and your data ownership is unclear, AI will surface all of that, faster and at greater scale than before. You’re not “adopting AI.” You’re giving your worst habits a faster processor.

Where Clean Data Pays Off

Netflix: Metacat as Infrastructure

Netflix published the technical details of Metacat on their Tech Blog in 2018. Metacat is a federated metadata access layer that gives engineering and data teams a unified view of distributed data assets, regardless of where they’re stored. It standardizes schemas, enforces ownership, and enables machine learning systems to be trained on verified, consistent inputs.

The operational context matters here. When Netflix expanded to 190-plus countries in January 2016, the scale demands on their content and data systems grew dramatically. Thousands of titles, hundreds of regional variations, multiple format requirements: none of that works without clean, governed metadata underneath it.

What Metacat demonstrates is that metadata infrastructure is a precondition, not a nice-to-have. The automated QC and ML-ready pipelines Netflix runs today are built on a foundation that took years and real engineering investment to establish. That foundation is the point.

Takeaway: standardized schemas and formal data ownership aren’t overhead. They’re the reason AI can do anything useful at scale.

Disney: AI Scene Tagging at Scale

Disney’s “Magic Words” tool, documented by Reuters in February 2024, uses AI and machine learning to scan its library for content, brands, imagery, and emotional tone. The output is structured metadata that advertising teams use for contextual targeting. AWS case studies have documented the broader scene-tagging program, led by Miquel Farré’s team, including deep learning models that identify animated characters across properties.

Disney’s library is one of the largest in entertainment. The only way to make it fully accessible to both internal teams and advertising partners is through machine-generated metadata that’s consistent and reliable enough to act on.

The key word is reliable. Character recognition and emotional tone scoring are only useful if the training data is clean, the models are validated, and the output is governed. Disney didn’t flip a switch. They built instrumented pipelines.

Takeaway: AI-driven metadata at this scale works because the underlying content library has structure that machines can learn from.

FOX Sports: AI in the Broadcast Booth

FOX Sports and Google Cloud built “FOX Foresight,” an AI platform deployed during the 2025 MLB World Series that gives on-air talent real-time data and storyline support during live broadcasts. The system surfaces relevant statistics, historical context, and narrative threads as games unfold, giving announcers a layer of analytical depth that would be impossible to produce manually in real time.

Live broadcast is one of the highest-stakes environments for operational process hygiene. There’s no margin for an AI system that pulls from inconsistent data sources or hallucinates a statistic on air. The success of FOX Foresight reflects disciplined work on data ingestion and validation upstream.

What Google Cloud and FOX built together is a system where the speed of AI is an advantage precisely because the data feeding it is trustworthy.

Takeaway: real-time AI in live production only works when the data architecture behind it is solid enough to be trusted under pressure.

BBC: AI for Open-Source Investigations

In 2023, the BBC used a combination of an internal facial recognition tool and a large language model to analyze social media posts and videos from the war in Ukraine. The investigation focused on a Vladivostok battalion and produced the “BBC Eye” documentary, which went on to win the Online Journalism Award for Excellence in AI Innovation.

This use case is different from the commercial applications above, but the underlying principle holds. The investigation required structured, validated data inputs: verified source material, documented methodology, and rigorous review. Cutting-edge technology. It required the same operational discipline as any well-run AI system.

The BBC didn’t hand a model raw, unverified footage and hope for the best. They built a process around it. The award they received reflects the quality of that process as much as the sophistication of the tooling.

Takeaway: governed data inputs and documented methodology matter whether you’re running ad targeting or award-winning investigative journalism.

Where Dirty Data Costs You

Unity: A $110M Hit From Bad Training Data

In May 2022, Unity disclosed that its Audience Pinpointer ad-targeting product had been compromised by bad training data from a large customer, compounded by a platform fault. CEO John Riccitiello attributed the problem directly on the earnings call. Unity cut its FY2022 revenue guidance by approximately $110M. The stock dropped roughly 36% the next day, wiping out around $5 billion in market cap, per Forbes reporting on May 20, 2022.

Audience Pinpointer’s job is to find the right users for a given advertiser at the right price. That job depends entirely on accurate behavioral and attribution data. When the training data was corrupted, the model’s targeting logic broke down, and it kept spending against bad signals until the damage was already done.

This is the automated-dysfunction scenario in its clearest form. The system worked exactly as designed. The problem was what it was given to work with.

Takeaway: an AI system that operates at scale on bad inputs doesn’t fail slowly. It fails fast and expensively.

Zillow: A $300M+ Algorithmic Misread

In Q3 2021, Zillow took a $304M inventory write-down on its Offers business and ultimately lost more than $500M as the program wound down. The company announced a 25% workforce reduction in November 2021, affecting approximately 2,000 employees.

Zillow’s algorithm was trained to value homes in a market that no longer existed by the time it was operating at scale. The post-pandemic housing market cooled rapidly, and the model consistently overvalued properties because it hadn’t been trained on anything like what was actually happening. The company was buying at prices that assumed stable or rising values, and the market was moving the other direction.

The failure wasn’t an exotic edge case. It was a foreseeable consequence of deploying a predictive model without adequate mechanisms to detect when its assumptions had stopped being true. No one was watching closely enough to catch the drift before it became a crisis.

Takeaway: AI models need ongoing validation against real-world conditions. A model that was right last year may be systematically wrong today.

McDonald’s: When the Pilot Doesn’t Scale

McDonald’s and IBM announced their AI-powered drive-thru voice ordering partnership in 2019. After more than two years of active testing at more than 100 locations, McDonald’s ended the partnership in July 2024. The company confirmed in June 2024 that it would turn off the technology at all test restaurants no later than July 26, 2024.

The public-facing record of the failure is unusually vivid. TikTok videos of the system producing bizarre and incorrect orders circulated widely, turning what was meant to be a seamless operational upgrade into a brand liability. The technology couldn’t handle the acoustic variability and menu complexity of real drive-thru environments at the volume and speed those locations require.

A controlled pilot that works is not the same as a production system that scales. The gap between the two is where operational process issues live.

Takeaway: piloting AI in controlled conditions and deploying it across a high-volume, high-variance operation are entirely different engineering problems.

Apple Intelligence: Summarization Without Guardrails

In December 2024, the BBC formally complained to Apple about Apple Intelligence notification summaries that were generating false headlines from real news notifications. The errors were specific and serious: one summary stated that Luigi Mangione, the suspect in the UnitedHealthcare CEO shooting, had shot himself. Another claimed darts player Luke Littler had won the PDC World Championship hours before the final was played. A third summary falsely stated that Rafael Nadal had come out as gay.

These weren’t editorial errors or bad data sources. They were hallucinations: the model was summarizing headlines and push notifications in ways that produced plausible-sounding but factually false statements. Apple disabled news and entertainment notification summaries in iOS 18.3, released in January 2025, and relaunched a revised version with iOS 26 in mid-2025.

The pattern here is an AI capability deployed into a high-visibility context before the failure modes were adequately understood or mitigated. Summarization is a genuinely hard problem. Deploying it on live news notifications, where a single hallucination reaches millions of people and damages credible third-party outlets, required guardrails that weren’t in place.

Takeaway: understanding how a model fails is as important as knowing what it can do. Deploying without that understanding is both an operational risk and a product risk.

The Pattern

Look across these eight examples and what surfaces is consistent.

Netflix, Disney, FOX, and the BBC all did significant work on data structure, ownership, and validation before AI could operate effectively on their assets. They invested in process before they invested in automation. The payoff is that AI works reliably, at scale, in production.

Unity, Zillow, McDonald’s, and Apple all ran into versions of the same problem: the AI system worked as designed, but the inputs, assumptions, or operational context weren’t ready for what the system would actually do. The failures weren’t random. They were predictable from the upstream conditions.

The pattern I notice across media supply chains specifically is that the metadata layer is where most organizations are underinvested. Not because people don’t know it matters, but because it’s hard to fund work that doesn’t ship a feature. As SVTA Metadata Working Group chair, I’ve seen this firsthand. The organizations that build formal metadata ownership and validation into their operational model before an AI initiative starts are the ones that can actually use AI when they need it.

The rest is just noise.

Before You Plug In

Before deploying AI anywhere in your media supply chain, an honest self-assessment of the following questions will tell you more than any vendor demo.

1. Are your workflows documented? Not in someone’s head, not in a Confluence page that hasn’t been touched in two years. Current, accurate, and owned.

2. Who owns your metadata? Not “the data team.” A named person or function with accountability for accuracy, completeness, and consistency.

3. How was your training data generated? If the answer involves manual processes with no QC, informal spreadsheets, or legacy systems with known quality issues, that’s what the model learns from.

4. What does failure look like, and how would you detect it? Unity’s system kept running on bad data. Zillow’s model kept buying overvalued homes. What’s your circuit breaker?

5. Is the pilot environment representative of production? McDonald’s learned this the hard way. If your test conditions don’t reflect real-world volume, variance, and edge cases, your pilot results don’t mean what you think they mean.

6. What happens when the model is wrong? Especially in any customer-facing or high-visibility context. Apple Intelligence failing on news summaries was a product problem and a brand crisis. What’s your rollback plan?

7. If AI scaled this exactly as it is, would we be proud of the result?

That last question is the one to sit with. The answer will tell you whether you’re ready.