A reader asked me: how do you manage underperforming peers, in a small team, with no leadership from above?

There are two things that make this hard, and they are hiding in plain sight. The first is in the word “manage”, and the trap here is that you can’t. A peer is not yours to manage. You have no authority over them, no ability to set their objectives, direct their work, or attach a consequence to anything they do or fail to do. The second part is “no real leadership from above”. This removes the escape hatch for “raise it with their manager”, when clearly there is vacuum where the manager should be. This is pretty much half of the actual problem.

I’m going to throw out the word “manage” here, because it’s just going to keep you attempting things you have no tools for, and replace it with things you can actually do.

That’s more or less the whole toolkit. Everything useful lives in there, and everything that will burn you out lives in trying to do the fourth thing, which is to be the leader that nobody appointed.

Subscribe now

First, work out which underperformance you’re dealing with.

Before you do anything, diagnose, because “underperforming” covers a multitude of problems that look identical from the outside but need opposite responses.

There is can’t: a genuine skill or clarity gap where the person would do better if they knew how, or knew what was actually wanted.

There is won’t: someone checked out, coasting, quietly deciding that the effort is not worth it.

There is blocked: someone drowning in something you cannot see - dependencies, overload, or a mess elsewhere.

There is drift: the entirely predictable result of a leadership vacuum, where there is no direction, no clear ownership, no deadlines that mean anything and no consequence for missing them. A normal person would slow to the level the environment rewards, which as it turns out, is not that much,

The last one matters more than the others, because if the real cause is the vacuum, then the person is not the problem and the absence of structure is, and no amount of helping the person will fix a thing that was never about them.

Supply the structure that’s missing, locally.

Confrontation is OUT. In a team with no leadership, the highest-leverage thing you can do is quietly become the source of the structure everyone is missing, without ever calling it that.

No leadership means no clear ownership, no visible commitments, no dates. You can introduce all three, not by decree, which you have no power to issue, but by proposal, which you do. “Shall we just write down who is taking which piece and roughly by when, so we stop treading on each other?” is not you managing anyone. It’s a reasonable colleague suggesting the team get organised, and almost nobody will say no to it because everyone is quietly suffering from its absence.

The trick here is that structure does what authority cannot. A peer will ignore you telling them to work harder. They will, however, find it much harder to ignore a shared and visible list, where they’ve publically taken a piece of work with a date on it, next to everyone else who has done the same. You’re making the absence of delivery visible, and letting the visibility apply the pressure - exactly the job the missing manager was supposed to do.

Make the work visible, because legibility is the manager you don’t have.

In a vacuum, underperfomance thrives because nothing is tracked. A simple shared record of what was committed and what was completed removes the fog. This also protects the person who is doing the work, probably you. The difference between the person carrying the team and the person coasting is invisible to everyone, including any leadership when it (eventually) arrives. Making it legible and the truth becomes self-evident without you having to say a single accusatory word, which is both kinder and far more effective than an argument.

Have the direct conversation, sideways.

You can, and often should, just talk to them. Not as their boss, because you aren’t, but as a colleague the work actually affects, because you are. The skill here is to make it specific, make it about impact rather than character, and bring the problem attached to a way forward.

Not “you need to pull your weight”, which is an attack of character they will defend against. Instead: “When the X piece lands late, I end up blocked on Y and we both miss the deadline. I would rather sort out what is getting in the way than hitting this. What’s going on with it?” That is a peer raising a real, shared, specific problem, and it opens the door rather than starting a fight. Sometimes, nobody has ever told them, plainly and without hostility, that it’s a problem. In a leaderless team the feedback loop that normally delivers that message does not exist. You can be it, once. What you cannot be is their performance manager on a permanent basis, so have the conversation, and then let it go rather than nagging. Nagging is just management without the authority to make it stick, and it’ll make you the villain,

Protect yourself, because their failure will try to become yours.

In a small team with no leadership, when things slip, the blame does not get allocated by anyone sensible, it usually just settles on who is most visible or most conscientious. Don’t let a peer’s underperformance quietly become your reputation.

First, decouple where you can: don’tt run your own critical path through the person who doesn’t deliver, if there’s any way to avoid it, because tying your outcomes to their reliability is volunteering to go down with them. Second, keep a neutral, unemotional record of dependencies and what happened, the same way you would document anything. Not to snitch, and not to build a case out of malice, but so that when the situation is eventually looked at, by a new manager (or anyone tbh), the account is accurate rather than whatever settled in the fog. In a vacuum, the person with the clear record of what actually happened is the person who does not get scapegoated.

The limit (which is the real answer).

The honest part (sorry!) because everything above will help and none of it will fully fix the situation. The problem, at its root, is the thing you were probably hoping was background: there is no leadership. You are being asked, mostly by yourself, to solve a leadership problem with peer toold, and peer tools only reach so far. You can supply structure, apply daylight, have the honest conversation, and protect yourself, and that will genuinely improve things. What you can’t do is be the permanent, unpaid, unacknowledged substitue for the management the team doesn’t have. The person holding it together while nobody notice, will burnout every single time, and their reward for holding it will just be more to hold.

Do the four things, and then watch what happens. The results will tell you what problem you actually have. If the team responds to a bit of structure and daylight, then the underperformance really was drift, and you have more or less fixed it. Congrats. You are the captain now. If it doesn’t, if the coasting continues, then the honest conclusion is that this is not fixable from your seat, because the missing ingredient is authority that you don’t have and can’t manufacture. The useful question stops being “how do I manage my peers” and becomes the harder one: how long am I willing to prop something up that’s not mine to prop up, in a place that has decided leadership is optional. That’s a stay-or-go question, and you should be asking it.

A reader sent me a situation this week that I suspect about forty percent of you are living in some version of, so, with their blessing, let’s get into it…

Someone senior in their chain, and, importantly, a kind and well-meaning person, requires every micro-decision to be cleared through them first. Because everything routes through one person, they are permanently stressed and unavailable. Because they are unavailable, the team waits days for answers. Because the team waits days, it misses deadlines (do you see where I’m going with this?). And because it misses deadlines, said Human concludes, with what feels like mounting evidence, that the team cannot be trusted to work without supervision and tightens their grip.

The reader called it a vicious circle, and asked how you break it.

I’d like to call out one thing before I really take the lid off: this loop contains no villain. This Human is not a tyrant. This is a kind person inside a machine that keeps manufacturing proof that they are needed. Every missed deadline arrives on their desk as confirmation of just that. They feel diligent. They feel like they are the responsible one. The worst bottlenecks in any company are almost never the power-hungry. They are the conscientious, drowning, genuinely helpful people who have mistaken being load-bearing for being useful, and who would be exceptionally hurt if you told them what they were costing you.

Subscribe now

What doesn’t work

Confrontation fails. Telling a conscientious Human “you micromanage and it’s hurting us” attacks their identity at a point I suspect they are most proud of. They don’t hear feedback, they hear ingratitude, from the very team whose missed deadlines they have been personally absorbing. You will get hurt feelings, a defensive tightening, and a Human who now believes the team doesn’t appreciate them. This is sub-optimal and will be worse than what you have now.

Waiting fails harder. Working around the delays, absorbing them, and hoping the habit burns out, simply feeds the machine. Every deadline that slips whilst you wait for sign-off, is logged by said Human as more proof. Patience in this loop is fuel. Say it with me: we don’t put fuel on fires.

Break the loop where it’s weakest

You cannot fix “stressed and unavailable”, that is their end of the loop. The end you hold is what happens while you wait, and what the Human sees when they look at the team. Let’s go…

1. Convert open-ended waiting into default forward motion.

The single highest-leverage change is to stop asking permission-shaped questions. A question like “how should we handle X?” hands them an open loop that sits in a queue for three weeks. Replace it with the decision made, plus an exit: “We are planning to do X for these reasons. Unless you flag concerns by Thursday, we will proceed.” You haven’t taken any control away. They can still say no, and this is precisely why it works on the kind Human: it is polite, transparent, and completely reasonable. But the default has flipped. Silence, which used to mean the team stalls, now means the team moves.

Do this gently and predictably, with real deadlines and real reasons, not as a trick. The first few times, pick decisions that are small and reversible, which brings us to point two.

2. Shrink what needs a decision at all, by agreement.

At a calm moment, not mid-crisis, someone the Human trusts proposes a sorting of decisions: the genuinely consequential ones, which absolutely come to them with a proper write-up, and the routine, reversible ones, which the team handles and simply logs where they can see. Frame it, honestly, as protecting their time for the decisions that deserve it: “We are wasting your judgement on font sizes, and would rather have it for where it matters”.

A conscientious person can hear that, because it flatters the diligence rather than attacking it. You are not asking them to let go. You are asking them to concentrate. Same outcome, entirely different feeling.

3. Replace the surveillance with a window.

The Human’s grip is soothed by visibility. Much of the checking is not really decision-making, it’s anxiety management, the Human reassuring themselves that things are on track. So give the reassurance before it’s sought: a short, regular, written update, what was decided, what is moving, what is coming to them and when. It feels like extra work but bear with me, it is actually the cheapest tranquilliser you will administer.

Every update that says “things proceeded without you and nothing caught fire” is a small dose of the counter-evidence that the loop has been starved of. You are retraining the belief, drip by drip, with receipts.

4. Make the cost of the bottleneck visible, kindly, in writing.

When a delay does hit, do not absorb it in silence and do not snipe about it next to the water cooler (yes, I know). Route it back, neutrally: “We were ready on 4th, received sign-off on 19th, which moved delivery to 26th.” No blame, no adjectives, just putting the information in a place it can be seen. This is the gentlest possible way of correcting the story the Human is telling themselves, because right now, in that story, the team misses deadlines. The record needs to show, calmly and repeatedly, that the team waits.

Over time, that reframing does more than any confrontation could, and if the situation ever needs escalating, the record already exists, and it was built without a single unkind word.

What breaking the loop actually looks like.

None of these things actually take anything from the Human. They flip a default, sort the queue, feed the anxiety with information instead of control, and correct the ledger. Together, they starve the loop of the only thing it eats - evidence that the team fails without them. When the record starts showing that the team deciding well, moving safely, and delivering on time whenever it isn’t waiting, the belief that drives the whole machine loses its food supply. Kind but controlling Humans, and this is the hopeful part, do actually update when the evidence changes, precisely because the control was never a malicious thing. It was just worry.

One thing: be honest with yourself about timsescales. Rome wasn’t built in a day, and neither was this loop. It will not unwind in a fortnight. You are running a slow campaign of counter-evidence, and the win condition is not a dramatic conversation where the Human sees the light. It is quieter: six months from now, the “unless we hear by Thursday” emails have become the culture, the routine decisions have stopped visiting their desk, and the Human is measurably less drowned, and if you have done it well, they believe things simply got better, which, in fairness, they did.

And if none of it moves, if every default gets overridden, every sorting agreement quietly reverts, every window ignored in favour of the grip, then you have learned something different and more important: the control is not worry, and no amount of evidence was ever going to be allowed to count. That is a different problem, with a different, more escalation-shaped answer, and knowing which problem you actually have is worth every bit of the six months it costs to find out.

I once cooked for a living in a kitchen that would not stay still.

I was a chef on a yacht, which sounds glamorous, and is, for the people being fed. For the person doing the feeding, it is ordinary cooking with one enormous complication, which is that the kitchen is at sea, and the sea does not care that you have a hollandaise going. The floor tilts. The pans want to leave the stove. Everything you set down slides, unless it is behind a fiddle rail, which is the little fence that stops your dinner throwing itself on the floor when the boat rolls. The stove itself is gimballed, hung so it swings and stays roughly level at the expense of your shins, while the whole room leans around it. You brace one hip against the counter and you keep working, because the guests will be on deck and they would like to eat at eight, and the weather is your problem, not theirs.

(I hope this is painting you a picture because I have been covered in sauce more than once).

Subscribe now

1. Stop waiting for flat water

You do not get a calm kitchen. You get the kitchen you get, on the sea you are on, and dinner still has to happen. If your standard of work relies on conditions being favourable, you do not have a standard, you have a preference.

Every workplace has its own weather: the reorg, the crisis, the strategy that changes every five minutes. And capable people essentially stop cooking, waiting for it to calm down so they can do their good work in peace. It does not calm down. Or it calms down long enough to lull you into a false sense of security, and then it rolls again. The people who deliver are the ones who stopped waiting for calm waters years ago.

2. Win the storm during the calm

Almost everything that saves you when it is rough was done when it was calm. In a kitchen, this has a name - mise en place. When the water is flat, you are not relaxing, you are prepping like the storm is coming, because it is: chopping, portioning, making the sauce, stowing everything into containers that close. So when it does get rough, you are not attempting delicate knife work on a pitching surface (this is a bad idea btw, ask me how I know). You are just assembling, from things already prepared, in the order you already know.

Under real pressure, you cannot improvise your way to quality. What carries you is what you set up beforehand: the documentation written when things were quiet, the relationships built before you needed them, the systems that make the crisis a matter of execution rather than invention. People admire the person who is calm in the storm and assume it is temperament. Mostly it is preparation. They are calm because they did this bit already, whilst everyone else was enjoying the sunshine.

3. Build for the roll you know is coming

You stow a galley assuming it is going to get thrown around. Nothing sits loose. Nothing is balanced in a way that only works when the room is level. You build for the motion in advance of it, so that when it does arrive, and it will, it is survivable rather than catastrophic.

The office version is to build in slack: buffer in the plan, redundancy in the system, margins that hold when something slips. A schedule with no buffer is a pan with no fiddle rail, fine until the first roll, and then everything is on the floor. You cannot see which wave will hit, you only know that one will. The people who plan to a hundred percent of flat-water capacity are always the ones clearing dinner off the floor.

4. Do not become a second storm

The galley is already in motion. The sea supplies all the chaos anyone could possibly need. The one thing that makes it unsurvivable is a cook who adds their own panic to the sea’s, moving fast and sharp and frightened, so that there are now two storms in a small room and only one of them was inevitable.

Your composure is the only thing in the situation you fully control. And it matters beyond you, becuase everyone under pressure is unconsciously calibrating off of the most senior person in the room, reading you to find out how frightened they should be. Stay slow and deliberate when the floor tilts and you give everyone permission to keep working. Flap, and you have taught the room that flapping is the correct response to weather, and now nobody is cooking.

5. Plate the calm

What goes up to the guests is a composed plate. They do not need to see you braced against the counter throwing fucks at a saucepan, and they should not have to, because absorbing the chaos so it does not degrade the result is most of what the job is. The report lands finished. The meeting runs clean. The team gets the decision, not the drama that produced it.

6. But never lie about the storm

The guests know there is weather. They are on the same boat. The chef is not pretending the sea is flat. The chef is declining to make the weather the guest’s dinner problem, which is craft. It is an entirely different thing than a captain announcing there is no storm whilst the boat fills with water, which is the lie that gets people hurt. Absorb the chaos so that it doesn’t ruin the output. Never absorb it so completely that the people who need to know a storm is coming are the last to find out. The first is professionalism. The second is how you drown politely.

7. Rest in the flat spells, because the sea is a cruel mistress

There will be nights you are seasick and exhausted and dinner is stil at eight and you make it happen anyway. That is real, and sometimes it is the job. But the chefs who last are not the ones who prove, repeatedly, that they can suffer. They are the ones who eat properly, sleep in the calm stretches, and provision the boat so that they are never one rought night from disaster. Endurance is not a personality. The person grinding through every storm on nothing is not tough, they are one bad week away from being no use to anyone, and the sea is cruel, and she always sends another storm.

The close

None of it makes the sea calmer, it just makes you the kind of person who can cook regardless, which, in a career that will not hold still any more than the sea does, turns out to be a pretty useful skill.

You bring someone a real problem. Not a vague anxiety, an actual, specific, this-is-going-to-hurt-us problem, and you have brought it because noticing it is literally your job. The person you bring it to is warm, capable, genuinely lovely to work with. They listen. They nod. And then they do something remarkable: they make the problem disappear without solving it. They reframe it, they reassure you, they wave at all the reasons it will probably be fine, and somewhere in there the conversation quietly stops being about the risk and starts being about you, and your energy, and whether you are perhaps a bit stressed at the moment.

You came in holding a problem. You left having been gently diagnosed as a bit negative. And the problem, for the record, is still there, sitting exactly where you left it, growing.

This is one of the quietly maddening dynamics of working life, and I want to be careful with it, because the easy version of this Substack article is a mean one, and the optimist, annoyingly, does not deserve a mean article.

Subscribe now

First, the case for the optimist

The positives are real and they are large, so let us do them properly. Optimists are, disproportionately, the people who build things. They take the risk the realist has talked themselves out of. They keep a team moving through the stretch where a more clear-eyed person would have quietly updated their CV. They see the opportunity rather than the forty reasons it will not work, and every so often they are right, and the thing exists because they believed it into existence while everyone else was being sensible about why it could not.

A team made entirely of pessimists, meanwhile, is a lovely, intelligent group of people who correctly identify every reason not to do the thing and then, with tremendous rigour, do not do the thing. You have been in that meeting. Everyone is right and nothing happens. It is like being trapped in a peer review of reality.

So this is not a piece about how to endure an exhausting person. It is a piece about how to work with a genuine strength that ships with one specific, predictable, occasionally maddening bug.

Why the bad news bounces

The optimist does not reject bad news because they are stupid or in denial. They reject it because, for them, optimism is not a mood, it is a working tool, the thing that manufactures the momentum and the morale that make them effective. And a problem, handed over raw, arrives as a threat to that tool. Acknowledging a risk feels, from the inside, dangerously close to giving up, which in the optimist’s private universe is the one unforgivable sin. So when you pass them a naked problem, they do not process it as information. They experience it as smoke drifting towards the alarm, and they respond the way people do at a slightly burnt dinner party. They open a window, wave a tea towel at it, and assure everyone the cooking is going brilliantly.

Once you see that, the whole thing stops being maddening and starts being solvable, because it tells you exactly what not to do. You are not going to win by crushing their optimism, and you should not want to, because you need it and so does the team. You are not going to convert them into a fellow pessimist, which would be a downgrade for everyone, yourself included, because one of you in the building is plenty. The skill is narrower and far more useful: deliver the hard news in a form their optimism can actually digest. You are not fighting the optimism. You are learning to smuggle facts into it.

How to be heard

Never hand them a problem without a plan attached. A naked problem reads as doom and slides straight off. A problem with a next step stapled to it is something they can hold, because it arrives pre-loaded with the one thing the optimist needs in order to look at bad news at all, which is a way forward. “This is at risk” bounces right off them. “This is at risk, and here is what I would do about it” lands. Every time. The plan is the spoonful of sugar, except the sugar is competence.

Frame it as protecting the goal, not threatening it. Optimists believe in the mission and are wired towards the win, so put the hard conversation in service of the thing they want rather than against it. Not “I do not think this launch is going to work,” which positions you in the doorway like the ghost of Christmas yet to come, but “I want us to hit this launch, which is exactly why we need to deal with this now.” You are standing beside them, pointing at the obstacle between them and the shiny thing. Same information, entirely different reception. It is a magic trick, frankly, and it works on approximately everyone.

Make it specific and evidenced, never a vibe. This is the one that protects you from the label. Vague pessimism, “I am just a bit worried about this,” is easy and honestly fair for an optimist to wave away as your temperament, because that is what it is. A precise, quantified, evidenced risk, “if the vendor does not sign by Friday the June date is gone, and here is the dependency that makes that true,” is a fact, and facts are much harder to reframe than feelings. Nobody has ever been called negative for reading out a date. The more specific the risk, the less it reads as your mood, and specificity is the entire antidote to being cast as the office raincloud.

Put it in writing, gently. Optimists genuinely forget the warnings, or remember them as having been much softer than they were, because optimist memory edits everything in the direction of it having been fine all along. It is not malice. It is the smoke tidying up after itself. So leave a light record, not as a gotcha, as a summary: aligning on the plan, flagging the risk, revisiting Friday. Two lines, friendly as anything, absolutely bulletproof. It protects you, and it means that when the thing happens, the record shows you were not hysterical, you were early, and there is no arguing with an email timestamp, however much everyone would like to.

Calibrate your alarm, or they will correctly learn to ignore it. This is the uncomfortable half, and it is on you. If everything is a crisis to you, the optimist is entirely right to tune you out, because an alarm that goes off constantly is not an alarm, it is a smoke detector with a dying battery, and everyone just learns to hate the noise. The credibility of your warnings is a budget. Spend it on the small stuff and you will have sod all left for the big one. Save the hard conversation for the risks that genuinely earn it, and become a worrier worth listening to, which is a rarer and more valuable creature than it sounds.

Give them the win too. You do not have to make them wrong in order to be heard, and you should not, because it triggers the exact defence you are trying to get around. There is almost always something genuinely going well. Name it, mean it, and then raise the risk. That is not arse-kissing. It is meeting them where they are, and it costs you nothing except the small private satisfaction of being maximally grim, which, be honest, you were enjoying a little.

When they still will not hear it

Sometimes you do all of this and it still does not land, and here you need a distinction that matters. There is a real difference between the person who struggles to hear bad news because they are genuinely, temperamentally hopeful, and the person who will not hear it because hearing it would oblige them to act, or to admit something they cannot afford to admit. The first is an optimist, and everything above works on them eventually. The second is not an optimist at all. They are using positivity as a shield, which is a darker and entirely different problem, and much closer to the person who insists everything is fine precisely because they need it to be. You know the type. This piece is not about them, and no amount of framing will fix them, because their problem was never the framing.

For the genuine optimist who still will not engage, the move is not to keep saying it louder, which only confirms you in the role of resident doom-monger, howling your predictions into the void. It is to build the safety net anyway, quietly: the buffer, the fallback, the documented risk. So that when the thing you could not get heard finally arrives, it is survivable, and you turn out to be the person who saw it coming, on the record, without ever once having become the office pessimist. You protect the work from the blind spot without making the optimist feel attacked. That, more or less, is the whole job, and nobody will ever thank you for it, which brings us to the end.

The close

The reason this is not a downbeat article is that you do not actually want to win this fight. You do not want to turn the optimist into someone who sees every risk you see, because a person who sees every risk builds absolutely nothing. You have worked with them too. They are worse company, they are less use, and their forecasts are just as wrong, only in the boring direction.

So you are not trying to fix them. You are trying to be the risk-radar bolted onto their momentum: the one person in the room quietly watching the things they cannot bear to look at, which is exactly what lets them keep looking forward. That is not a smaller role than theirs. It is one of the most valuable positions an operator can hold, and it is almost entirely invisible, and it works precisely to the degree that you never once get to say I told you so.

Well. Almost never. You are allowed one, per career, and you should choose it with enormous care and enjoy it completely.

First, the one law that governs the whole document, because once you clock it you cannot unclock it. A company does not put a value on the wall because it has that value. It puts it there because it is quietly shitting itself that it does not. You never need to remind people to do the thing they already reliably do. Nobody has “we breathe” on the wall. So the values page is not a mirror, it is an X-ray, and every single line on it is the company grassing itself up about the exact spot where it knows, somewhere deep down, that it is falling apart. Read like that, it becomes the most honest thing the company will ever produce, entirely by accident and completely against its will.

There are always five.

“One Team”

What it says: something about being one team, pulling in the same direction, winning together and, inevitably, learning together.

What it means: we are four teams who do not speak to each other but do share a Slack workspace, and someone at an offsite decided the fix for that structural problem was a noun. “We win together” means the credit for anything good floats gently upward and out of your reach. “We learn together” means that when it all goes tits up, it was, on reflection, a collective learning experience, and therefore nobody in particular has to own the fuck-up. Marvellous system. Very calming for everyone except the person who actually did the work.

“Excellence”

What it says: something about setting a high bar and holding ourselves to it in everything we do.

What it means: absolutely nothing, and that is the beauty of it. “Excellence” is the plain rice of the values deck, the one you reach for when you cannot think of a real value but the slide needs a fifth bullet. Excellent at what. Measured how. Against whom. It commits to fuck all and can therefore never be broken, which is precisely why it feels so safe to write down. And “we hold a high bar” is a phrase spoken exclusively by companies in the enthusiastic process of lowering one, usually mid hiring-spree. A bar you are genuinely holding does not need a poster. People can just feel it in the room, the same way they can feel its absence.

“Built for the Real World”

What it says: something about caring what happens out in the real world rather than in a controlled demo, because if it does not work where it counts, it does not count.

What it means: we ship a truly heroic quantity of demos. You do not write “not demos” on your own wall unless the demos are, structurally, the entire product. It is the corporate equivalent of a hand-painted sign reading “this is not a scam,” which is a sign that has only ever, in the history of the world, appeared on a scam. The companies actually having real-world impact are, as a rule, far too busy having it to stop and carve it into a value.

“Transparency”

What it says: something about being open and direct with each other, because that is how real trust gets built.

What it means: right, this is the darkest and most reliable law in the whole genre, so tattoo it somewhere. The presence of “transparency” on the values page is inversely correlated with the presence of any actual transparency in the building. A genuinely honest company does not need to remind itself, in writing, next to the mission statement, to be honest, in the same way an honest bloke does not open every sentence with “to be totally straight with you.” The reminder is the tell. If the people at the top have felt the need to formally, publicly commit to being direct with you, congratulations, you have your answer, and the correct response is to start reading everything they say exactly twice. It is not a promise. It is damage control with a nice font.

“Ownership. Bias for Action.”

What it says: something about taking ownership of outcomes rather than tasks, moving fast, and treating the whole thing as ours to win.

What it means: “ownership” is the warm, flattering word your company uses for the enormous pile of unpaid scope it is about to bung directly onto you and call a compliment. “Bias for action” means please, for the love of god, stop asking clarifying questions and stop waiting for the decisions that are never, ever coming from upstairs, and simply crack on, so that when the thing breaks it is unambiguously your name attached to it and not the name of anyone who could actually have prevented it. “We move fast” you already understand. And “this is ours to win” is the specific, plucky, faintly manic little noise a company makes at the exact moment it has, quietly, started to lose.

The blank page at the end

Stack all five together and you have not learned a single thing about what the company is. You have learned, in its own handwriting, precisely what it is terrified it is not: not aligned, not rigorous, not real, not honest, and not decisive, roughly in that order, or it would not have needed the bloody poster. The values page is the company grassing itself up, which is why it is worth reading a great deal more closely than the marketing it thinks it is.

Which means the most honest values page you will ever read is the one that does not exist at all, produced by the rare and slightly unnerving company that is simply too busy being aligned, rigorous, real, honest and decisive to get round to informing you, in a serif font, that it is. You will know that company when you find it. Not from anything on its wall, but from the deafening absence of one.

What a technical operator actually does

A technical operator is not the person doing the deepest technical work in the room. They are the person who makes a group of people and systems reliably produce outcomes. The engineers write the code, the researchers run the experiments, the designers design. The operator makes sure the right things get decided, the right people are unblocked, the handoffs do not get dropped, and the thing that was going to quietly fail gets caught while it is still cheap to catch.

If you want the whole job in one idea: your core skill is entropy reduction. Left alone, any organisation drifts towards more confusion, more dropped threads, more decisions that nobody quite made, more surprises that nobody saw coming. Your job is to push the other way. You turn ambiguity into decisions, scattered people into coordinated ones, and surprises into survivable events. Everything in this manual is an application of that single idea.

One warning that the rest of the manual keeps coming back to: this work is invisible when you do it well. A meeting that runs cleanly, a dependency that did not blow up, a decision that got made on time, none of these leave a mark. That invisibility is the job, and it is also the trap, and the final section is about not falling into it.

Read more

Someone on your team will come to you, quietly, slightly too casually, and ask some version of the question. Should I be worried. They will phrase it as though it is a small question, and you will both know it is not, and in the half second before you answer you will feel the entire weight of the situation arrive, because whatever you say next is going to matter more than almost anything you say to them all year.

That moment is what I want to talk about. Not the transition itself, the leadership departures and the resignations and the strategy that changes every fortnight, because you did not cause those and you mostly cannot fix them. This is about the far narrower and more answerable question of what you do for the people around you while it is happening, from the middle, where most of us actually sit, with responsibility for a team and no authority over the thing that is frightening them.

Most advice about leading through change is written for the person at the top, and it is about vision, and it is useless to you, because in a real transition the vision is precisely the thing that just walked out of the building. You do not have a grand reassuring narrative to offer. What you have is something smaller and, it turns out, far more valuable, if you are willing to do the one thing the situation will pressure you hardest not to do.

Subscribe now

The one thing you must not do

You must not reassure them.

Not in the way the instinct, and the people above you, will push you to. When an organisation is coming apart, a specific instruction tends to flow downward, sometimes explicit and sometimes just in the air: keep everyone calm, keep them focused, tell them it is fine. And the temptation to comply is enormous, because you care about them and because a frightened team is harder to lead and because “everything is going to be okay” is the sentence that makes the immediate discomfort go away.

It is also the sentence that ends your usefulness to them, instantly and permanently. Because they can see that it is not fine. They are not stupid. They have watched the same people leave that you have. So when you tell them everything is fine, you have not comforted them, you have told them something much worse than the truth, which is that you are now one of the people who will manage them rather than level with them. You have moved yourself, in one sentence, from the short list of people they can trust to the long list of people they have to decode. And in a transition, a trusted source of truth is the single scarcest and most valuable thing on offer, and you have just resigned the post.

The entire foundation of steadying a team is this: you can decline to tell them things, and you can tell them you do not know things, but you can never tell them something false. Get that one rule right and everything else works. Get it wrong and nothing else you do will matter, because it will all be filtered through the knowledge that you are performing rather than telling.

What a transition actually takes from people

It helps to know what you are fighting, because a transition does its damage through three specific mechanisms, and you can push back on all three locally even when you cannot touch them globally.

It takes clarity. Suddenly nobody knows what is happening, what the plan is, who is in charge, or what next quarter looks like, and the human mind abhors that vacuum and fills it, reliably, with the worst available story. It takes safety. The unspoken question under everything becomes am I next, is this sinking, should I get out, and a person asking that question several times an hour is not a person doing their best work or sleeping well. And it takes momentum. When the future is fog, the natural response is to stop rowing, because why pour yourself into work that might not matter for a company that might not be here.

You cannot restore clarity, safety, or momentum at the level of the whole organisation. That is above your pay grade and, in a genuine crisis, possibly above everyone’s. But you can manufacture a local version of all three for your specific team, and that local stability is simultaneously the most useful thing you can do and the honest limit of what you can promise.

Restoring clarity: shrink the unknown to its real size

Left alone, the unknown feels infinite, and infinite is terrifying. Your job is to cut it down to its actual, survivable size.

Tell them what you genuinely know, plainly. Tell them what is decided and what is still open. Above all, tell them what is not changing, because in the noise people lose sight of the large amount that is still stable: the work in front of them this week, their pay this month, the next thing you are shipping, the fact that they still have a job today. Name the parts of the fog you cannot see into, and say when you expect to know more, and then actually come back when you said you would. “This is what I know, this is what I do not, this is when I will have more” is a sentence that does more to calm a team than any amount of optimism, because it treats them as adults and it makes the unknown finite.

Restoring safety: be predictable when nothing else is

Safety in a transition does not come from promising people they are safe, which you cannot honestly do. It comes from you being a stable, predictable presence when everything above you is lurching. The one-to-one still happens, at the normal time, and you are the same person in it that you were last month. You do not catastrophise at them, and you do not perform a serenity you do not feel either. You keep your commitments, small and large, because every kept commitment is evidence that at least this one part of their world still works the way it is supposed to.

And you become a filter, not an amplifier. The chaos above will generate an enormous amount of noise, panic, contradictory signals, doom, and a great deal of it will reach you. Do not simply pass it through. Absorb what you can, and transmit downward only what is real and actionable. Protecting your team’s attention from the thrash is not keeping secrets from them; it is refusing to make them live inside every tremor when only some of the tremors are theirs to feel.

Restoring momentum: give them something solid and near

When the long term is unknowable, do not try to motivate people with it, because you will be asking them to draw energy from the exact thing that is frightening them. Give them something solid and close instead. A concrete goal for the next two weeks. A problem worth solving that is real regardless of what happens to the company. The ordinary satisfaction of doing good work with people they like, which is a genuine and durable source of meaning even when the surrounding structure is in question. You are not lying to them about the future. You are giving them a floor to stand on while the future sorts itself out, and a floor is worth a great deal when you cannot see the ceiling.

The two harder things

Two more, and they are the ones that separate steadying a team from merely reassuring one.

Handle the people who leave well, because everyone who stays is watching how you do it. People will resign; in a real exodus, good ones will. Do not guilt them, do not take it personally, do not badmouth them once they have gone, and do not silently punish the people who remain by dumping the departed’s work onto them as though it were free. How you treat the leavers is read, closely, by everyone staying, because it is a live demonstration of exactly how they will be treated when it is their turn, which some of them are already privately certain it will be. Generosity towards the people walking out is one of the loudest signals of safety you can send to the people still at their desks.

Tell them the truth about their options, even the option where they leave. This is the bravest thing on the list and the one a loyal company person will tell you not to do. If someone on your team asks, honestly, whether they should be keeping an eye on the market, the true answer, in a company that is genuinely coming apart, is often that yes, it is entirely reasonable to keep your options open, and that you will not think less of them for it. Saying this feels like a betrayal of the org. It is not. Your duty of care is to the human being in front of you, not to the fiction that everything is fine, and a person who later discovers you steered them into staying on a sinking ship to keep your own team intact will never trust you again, and will be right not to. You can want them to stay and still tell them the truth that would let them go. Do the second one.

A note on your own oxygen

You cannot be the steady floor for other people if you are quietly falling through it yourself. Being the calm one for a whole team while the ground moves is a real and depleting job, and it comes with the specific loneliness of being unable to fall apart in front of the people relying on you not to. So you need your own version of what you are giving them: someone to be honest with, off the stage, who is not on your team and not on the cap table. Steadying others is not the same as being fine, and pretending otherwise is how the person holding everyone together becomes the last one to notice they have stopped being okay.

Finally,

You cannot save the company from where you sit. It is worth being clear-eyed about that, because a great deal of pointless suffering comes from good operators taking personal responsibility for outcomes that were decided several floors above them. What you can do is much smaller and genuinely worth doing: you can be the reason a handful of people got through a frightening stretch with their nerves and their trust in at least one person intact, told the truth the whole way, and came out the other side either still standing where they were or gone somewhere better with your blessing.

That is not fixing it. It was never in your power to fix it. But being the one steady, honest thing in a season of instability and dishonesty is not a consolation prize. For the people on your team, it may be the only part of the whole experience they will remember without flinching, and if it comes to it, the same honesty you owe them is the honesty you owe yourself about when it is time to go too.

1. They narrate decisions, not activity

The junior update is a list of activity: I looked into the thing, then I checked the other thing, then I had a call about it. The senior update is a decision: I’ve decided we go with A, here is the one reason, here is what I need from you. Same work underneath, completely different signal. Activity says “here is proof I was busy.” A decision says “you can hand me a problem and get back a resolution, not a status report.” Start reporting the conclusion and the ask, not the journey, and watch how differently people respond to you.

2. They say less, and later

The most senior person in the room is very often the quietest for the first ten minutes. This is not shyness. Speaking first spends your credibility on the least-informed version of your opinion, before the actual problem has surfaced. So they let it emerge, and then say one thing that moves the conversation, rather than six things that decorate it. The amount someone talks in a meeting is usually inversely related to how much each sentence they say is worth.

3. They name the trade-off, not the answer

Asked “what should we do,” the junior instinct is to produce an answer, which can then be wrong. The senior move is to hand back the trade-off: if we do A we get speed but we carry this risk; if reliability matters more than the date, it’s B. This is more useful to the person asking, it is far harder to be wrong about, and it quietly signals that you can see the whole board rather than just the move in front of you. Answers make you a contributor. Trade-offs make you someone consulted before the decision.

4. They volunteer the uncomfortable number first

The junior handling of bad news is to soften it, delay it, or wait to be asked. The senior handling is to say it first, plainly, with the plan already attached: we are two weeks behind, this is why, this is what I am doing about it. Volunteering the bad number before anyone has to drag it out of you is one of the highest-trust moves available, because the single most valuable thing a manager wants to know about you is that you are not the person quietly sitting on the problem that later explodes. Say the number out loud. It costs you a bad afternoon and buys you years of being believed.

5. They stop asking permission for reversible things

Juniors ask permission to offload their own risk: is it okay if I do X. Seniors inform rather than ask for anything reversible, and reserve the actual asking for things that genuinely cannot be undone. “I’m going to do X on Thursday unless you see a problem” instead of “would it be alright if I did X.” It signals that you can be trusted with judgement, and it stops turning your manager into the bottleneck for trivia, which they resent more than they will ever tell you. Ask for the irreversible. Announce the rest.

6. They make other people look good on purpose

Counterintuitive, but the people who read as most senior are frequently the ones handing credit outward, building openly on someone else’s point, making the room smarter rather than competing for airtime in it. Insecurity hoards; standing distributes. Naming whose idea it actually was, and adding to it rather than talking over it, looks like generosity and functions as a status move, because only people secure in their own position can afford to spend credit rather than collect it. The person who makes everyone else sharper is the one the room starts organising itself around.

7. They can say “I don’t know” without flinching

The junior fear is that a visible gap in your knowledge exposes you as a fraud, so juniors perform certainty they do not have. The senior version is “I don’t know, here is how I’d find out, I’ll have it by Thursday.” Certainty theatre is a tell. Calibrated confidence, which means knowing and stating the edges of what you actually know, is the opposite tell. Admitting the gap signals standing precisely because only someone reasonably secure in their standing can afford to admit it. The flinch is the thing that reads as junior, not the not-knowing.

8. They protect the room’s time

Juniors fill time: the long update, the over-explanation, the covering of every base in case they are asked. Seniors compress. The one-line summary. The “we don’t actually need to solve this here, let me take it offline.” The killing of their own tangent halfway through because it stopped being useful. Treating everyone’s attention as the expensive, finite thing it is signals that you operate at a level where attention, not effort, is the scarce resource. The person who reliably gives a meeting its ten minutes back is remembered as competent long after anyone forgets what was decided.

One thing worth saying…

Several of these moves are not read the same way on every person, and it would be dishonest to pretend otherwise. Going quiet reads as gravitas on some people and as disengagement on others. “I don’t know” lands as secure calibration for some and gets logged as a competence doubt for others. Crediting the room is leadership when certain people do it and invisibility when others do. The moves still work, and they are still worth doing, but if you are someone the room does not already code as senior by default, you will often have to do them a touch more deliberately, and pair them with the receipts, to get the same reading a more default-senior colleague gets for free. That is not a reason to skip them. It is a reason to know what you are working with.

The through-line

None of these is about sounding clever. Every one is a way of absorbing uncertainty rather than generating it: handing back decisions instead of activity, trade-offs instead of guesses, bad news instead of silence, compression instead of noise.

Seniority, stripped of the mythology, is mostly the reputation for making the people around you less anxious about the thing you are responsible for.

Do that reliably enough and the title tends to arrive as a formality, slightly after everyone has already started treating you as though you had it.

There is a tab you have closed. I have closed it too, so I am not going to pretend this is a lecture from higher ground.

You found the role. The good one, the one that made something sit up in your chest, the one you read twice. And then you got to the requirements, and there was the bullet you do not have. Eight years and you have six. The specific tool you have never touched. The degree in the wrong subject. One line out of a list of twelve that you cannot honestly tick. And you closed the tab, and you told yourself you were being sensible, realistic, not wasting anyone’s time, and the sensible feeling was so convincing that you did not notice you had just declined a job nobody had refused you.

I want to talk about that tab, and about the story we have been told to explain it, because the story is wrong in a way that matters, and the truth underneath it is far more useful and much harder to say on a motivational poster.

The stat you have heard, and the part you have not

You have almost certainly heard the figure. Men apply for a job when they meet about sixty per cent of the qualifications; women apply only when they meet a hundred. It gets quoted everywhere, in the big career books, in the LinkedIn posts, always with the same tidy moral attached: women simply need more confidence. Believe in yourself. Have more faith. Lean in.

Two things about that figure are worth knowing. The first is that its origins are a great deal flimsier than its fame suggests. It traces back to an old internal company report that was never published, never peer-reviewed, and by some accounts was not really a study at all but a single executive’s offhand answer to a question. It has been repeated so many times that it has hardened into a fact, which is a process you should always find suspicious, because it is exactly how comfortable stories survive without ever being checked.

The second thing is the one nobody puts on the poster. When a researcher actually went and asked over a thousand people why they had not applied for jobs they wanted, the answer was not a lack of confidence. Barely anyone said they held back because they doubted they could do the work. The most common reason, for women and men both, was some version of this: I didn’t think they’d hire me if I didn’t meet the listed qualifications, and I didn’t want to waste my time. Not “I can’t do it.” “They won’t pick me.” Those are completely different problems, and we have spent a decade solving the wrong one.

The actual gap is not confidence

What that survey exposed is not a confidence gap. It is a rules-literalism gap, and it is far more interesting.

The people who did not apply were not cowering. They were being obedient. They read the list of requirements and did the reasonable, rule-following thing, which was to believe that the requirements were, you know, required. That the posting was a contract, and the qualifications were the entry fee, and turning up without all of them was a bit like sitting an exam you had not revised for. They took the document at its word.

The people who did apply had understood something the rule-followers had not: that a job description is not a contract. It is a wish list. It is a document a busy person wrote in twenty minutes, describing the mythical candidate who does not exist, and it is understood by everyone on the hiring side to be aspirational. Hiring does not run on checklist completion. It runs on advocacy, on relationships, on how the work is framed, on whether someone in the room wants you, on a dozen human things that a bullet-pointed list cannot capture and was never meant to. The confident applier was not braver. They had simply, correctly, read the wish list as a wish list, and treated its contents as strong suggestions rather than locked doors.

This reframes what “self-worth” even means here. You have been tying your sense of whether you are good enough to a document that was never an assessment of whether you are good enough. It was a recruiter’s wish, written quickly, optimistically, and slightly carelessly, and you handed it the authority to tell you what you are worth. Stop doing that. Not because you should love yourself more, but because you are outsourcing your self-assessment to a spec sheet that even the people who wrote it do not take literally.

Who gets taught the rules are real

This is where it stops being universal and becomes the specific thing worth talking about, because rule-literalism is not distributed evenly, and it is not an accident who ends up holding it.

Some people are raised and rewarded, from very early, for treating rules as binding: follow the instructions, meet the criteria, do not put yourself forward until you are certain, do not make anyone’s time your business unless you have earned it. Other people are quietly permitted, from equally early, to treat rules as negotiable, to ask for the thing they have not obviously earned, to read a “required” as a “preferred” and back themselves anyway. Women, disproportionately, get taught the first version. So do plenty of people who were told, in one way or another, that they would have to be twice as qualified to be considered half as seriously, which is advice that is both true and quietly ruinous, because a person who believes they must be twice as qualified will apply for half as much.

The confident sixty-per-cent applier, in other words, is not more talented than the woman reading the same posting and closing the tab. He has simply been operating, his whole career, on the correct information about how binding the rules are, information she was actively discouraged from having. The fix for that is not to tell her to have more confidence, which is a bit like telling someone to be taller. The fix is to give her the information: the rules were always suggestions, you were misinformed on purpose, and the person applying over your head knows something you were kept from knowing rather than possessing something you lack.

The worst they can say

Which brings us to the line everyone reaches for, “what’s the worst they can say, no?”, and I want to rescue it from the pep-talk register and put it back where it belongs, in arithmetic.

The reason to apply is not that you should feel brave. It is that the sums are lopsided in a way that makes holding back irrational. An application costs you almost nothing: a rejection you will have forgotten by the following week, from people you will likely never meet, about a decision that reveals nothing about your worth and everything about a process you cannot see. That is the entire downside. The upside is the job. And the one option on the board that guarantees you do not get it, with total certainty, every single time, is the application you do not send. A no from them is a maybe you gave yourself a shot at. The only reliable no in the entire system is the one you issue to yourself, quietly, by closing the tab and calling it realism.

You are not sparing anyone your time. There is no queue you are being polite by staying out of. The recruiter does not thank you for pre-rejecting yourself; they simply never know you existed, which is the precise outcome you were afraid of, arrived at voluntarily and for free.

The close

I am not going to tell you to believe in yourself, because that is the advice that has been failing at this for a decade, and it fails because it treats a problem of information as a problem of feeling. You do not have a shortage of self-esteem. You have been given a document and told it was a test, when it was only ever a wish, and you have been reading it far more honestly than the people who wrote it ever intended.

So apply for the one you do not quite qualify for, not as an act of courage, but as a correction. Read the requirements as the suggestions they always were. Frame the six years as the strength they actually are and let the person on the other side decide, which is their job, not yours to do for them in advance. The belief that you will land the next job you want is not optimism, and it is not confidence, and it is certainly not manifestation. It is an accurate reading of a game you were taught to take too literally, by people who were quietly playing it looser the whole time.

The job you talked yourself out of is still there. You were never underqualified for it. You were just better at following the rules than the rules deserved.

(Thanks for bearing the silence last week btw, I took my own advice and went on holiday. 10/10 highly recommend)

1. An option is not a share, and a share is not money

Most startup “equity” is not equity yet. It is options: the right to buy shares later, at a fixed price, if you choose to. Until you exercise that right you own precisely nothing, and even once you do, you own shares in a private company, which is not money. It is a claim on money that may exist one day, if a specific and not very likely sequence of events occurs. The word “equity” is doing an enormous amount of quiet lifting in that offer letter, and most of what it lifts is your enthusiasm.

2. The strike price can quietly sink the whole thing

To turn your options into shares, you pay the strike price, the per-share price fixed when they were granted. If the company is later worth more than your strike, good. If it is worth less, your options are “underwater,” which is a gentle way of saying worthless, because nobody exercises the right to buy a thing for more than it is worth. Ask what your strike price is and what the company was most recently valued at. If the recruiter cannot tell you both, that is itself the answer to a different question.

3. The cliff means a year of nothing

Equity almost always vests over time, typically four years, with a one-year “cliff” at the start. Leave, or get pushed, before that first year is up, and you walk away with zero, regardless of the impressive number in the offer. The cliff is reasonable in principle and brutal in a year when companies are cutting staff at ten months to save exactly this. Please please please know your cliff date.

4. The exercise window when you leave (it’s a trap!)

This is the one that ambushes people, so read it twice. When you leave a company, you typically get a short window to exercise your vested options, often around ninety days, though it varies by company and country. Exercising means buying the shares, in actual cash, out of your own pocket, sometimes a five-figure sum, and possibly triggering a tax bill on top. Miss the window and the equity you spent years earning simply evaporates. Enormous numbers of people discover, at the worst possible moment, that their “equity” requires them to write a large cheque to keep it. Ask, before you join, whether the company offers an extended exercise window. Some do, and these are the ones that will happily tell you.

5. Your slice shrinks every time they raise

Every funding round issues new shares, which means your percentage gets smaller, a process with the honest name of dilution. Your 0.4 percent today is not 0.4 percent at exit; it is 0.4 percent minus every round between now and then. This is not automatically bad, a smaller slice of a much larger pie can still be worth more, but it is real, it is relentless, and nobody sends a memo when it happens. Assume the number you were quoted is the most flattering it will ever be.

6. You are last in the queue, behind the investors

Investors hold preferred shares. You hold common. The difference is not cosmetic. Preferred shares come with liquidation preferences, which means investors get their money back, often one times their investment, sometimes more, sometimes doubly through something charmingly called participation, before common shareholders see a single penny. A company can sell for a number that makes a headline and leave employees with almost nothing, because the preference stack ate the proceeds first. Ask how much has been raised and on what preference terms. It is the difference between a life-changing exit and a commemorative mug. Ask me how I know.

7. A percentage is meaningless without the denominator

“You will get 0.5 percent” sounds like something. It is nothing until you know the total number of shares and the current valuation, which together are the only way to turn a percentage into a number of actual pounds or dollars. Ask for your number of shares, the total shares outstanding, and the latest valuation. Anyone genuinely offering you a good deal will not mind you doing arithmetic. The reluctance to give you the denominator is the most informative part of the conversation.

8. Tax can arrive before the money does

Depending on where you live and the type of equity you hold, the act of exercising or vesting can itself be a taxable event, meaning you can owe tax on paper gains you cannot sell anything to pay for. I am going to be plain here, because this is the one place in this list where getting it wrong is genuinely expensive: I am not a financial or tax adviser, the rules differ enormously by country and by scheme, and this is the rare situation where an hour with someone who does this for a living pays for itself several times over. Before you sign anything, ask what tax events your equity creates, and when they land.

9. Most of it is worth nothing, and everyone selling it to you knows

The base rate is the thing the excitement in the room is designed to stop you calculating. Most startups fail, and many of the ones that do not fail exit modestly enough that the preference stack absorbs the lot. The expected value of most startup equity, honestly computed, is low. This does not mean never take it. It means treat it as a lottery ticket, pleasant as upside, ruinous as the reason you accepted below-market cash. The cash is the only part of the offer that is unambiguously real. Negotiate that as though the equity were worth nothing, because statistically, it usually is.

The through-line

Notice that every item on this list is a question you could ask and mostly will not, because asking feels ungrateful, or unsophisticated, or as though you are the only person in the room who does not already understand. You are not. Almost nobody understands, the complexity is doing exactly what it was built to do, and a component you cannot value is one you will reliably undervalue in their favour and overvalue in your own hopes.

So ask the boring questions. Get the answers in writing. Value the cash, because the cash is real, and treat the lottery ticket as a lottery ticket, which is to say: buy it cheerfully, hope quietly, and never, ever remortgage the house on it.

Your equity might change your life. It is simply that the odds are printed in ink too small to read, on purpose, and the person reading it to you is being paid to sound excited.

There is a specific moment, at a company that is quietly coming apart, when you realise everyone in the room is looking at you.

Not because you have said anything. Because you are the one who has not started shouting, and in a room where everyone else has, the person still visibly calm becomes the instrument the others read to work out how frightened they are allowed to be. You are the dial. And the second you understand that you are the dial, you also understand that you are no longer permitted to move, because if the dial swings, the whole room swings with it.

I have been that dial, recently enough that I can still feel it, though this was a job or two ago and I am going to keep the details deliberately blurred, because the details belong to specific people and the pattern, unfortunately, belongs to everyone.

Subscribe

The weather

The company had promised its clients something that did not exist. Not “was running behind.” Did not exist. Someone had flogged the clients a product that lived exclusively in a slide deck, on the serene assumption that it could be conjured in the short stretch between the promise and the delivery date, which is precisely the assumption made by people who will not personally be doing any of the conjuring. So a small number of us were handed a deadline that was really just someone else’s optimism with a calendar invite attached, and instructed to go and make it retroactively true.

On its own, that is a normal bad quarter. But it arrived, as these things always do, with weather. The team had turned on itself in the specific way teams do when they are frightened and under-resourced and hunting for somewhere to put the blame that is not the ceiling. And there was one person doing a genuinely heroic amount of the blaming: loud, fluent, endlessly gesturing at a particular colleague as the blocker, the reason it was all so hard, the rot at the centre of the thing.

It was him, obviously. When anyone finally bothered to look, the loud one was the problem, and not subtly. He was the person who had sold the clients the imaginary product in the first place, and the entire performance of blaming a colleague was just an early and energetic effort to arrange for somebody else to be holding the bag at the exact moment everyone noticed it was empty. The villain was not lurking in the shadows being quietly menacing. The villain was standing in the middle of the room shouting “villain” and pointing, with a brass neck you could almost admire right up until it swung round to you. And it very nearly worked, because a person confident enough to accuse loudly is, in most organisations, believed by default over a person calm enough to just be getting on with the job. Volume reads as veracity. It bloody shouldn’t, but it reliably does.

Presiding over all of it was the founder, trying with enormous, genuine sincerity to make everybody happy, which remains the single most reliable method ever devised for making everybody furious. Every call was a compromise engineered to offend no one, and so every call offended everyone, a slow shambles of consensus in which each decision was built so carefully to dodge a hard choice that you could see the exact shape of the spine it was avoiding. And a compromise that pleases nobody still has to be carried out by people, and the people could all tell precisely what it was.

And then the person I reported to came apart. I will not say how, because it was human and it was theirs and they did not sign up to appear in anything I write. But it happened, and in the quiet afterwards there was a small, wordless redistribution of everything they had been holding. Most of it landed on me. Nobody decided this out loud. It simply flowed downhill, and I was the nearest flat surface.

The designated adult

Which is how I became the designated adult.

Nobody appoints the designated adult. There is no meeting, no title, no line in a document. It is an emergent property of a system under stress, and the mechanism is this: in a crisis, the person who stays functional becomes the place all the dysfunction goes to be processed. Not because it is fair. Because it is efficient, and panicking organisations optimise for efficiency with a brutality that calm ones only ever talk about. You did not fall apart, so you are handed the things that are falling apart. You kept your head, so you are lent everyone else’s. The reward for being able to carry it is more of it, and the reward for carrying that is more still, in a loop with no natural exit, because the load is not calibrated to what is reasonable. It is calibrated to whatever you have not yet visibly failed to hold.

This is the competence tax, and it is the most quietly punishing arrangement in any workplace, because unlike ordinary exploitation it feels, from the inside, like being trusted. They are relying on you. They say so constantly. And relied-upon sits so close to valued that it takes an embarrassingly long time to notice they are not the same word, and that only one of them ever turns up on a payslip. The other turns up solely in your inbox, at volume, at all hours, with a cheerful little flag on it.

It is not handed out at random

I would love to tell you the designated adult is chosen by lottery, that this rota of unpaid emotional stewardship rotates fairly around the team. It does not, and you already know it does not, because you can almost certainly picture the designated adult at your own workplace, and I would bet the price of a subscription on which way that picture leans.

The behaviours that quietly nominate you for the role, noticing the thing about to fall through the gap, tidying it before it becomes a crisis, managing everyone’s feelings, remembering what is due and who is upset and where all the bodies are buried, being the reasonable one so that at least someone is, are precisely the behaviours women are raised to perform and then expected, at work, to supply for nothing. It is office housework. And like housework it is invisible until it stops, thankless while it continues, and somehow never quite anybody’s actual job, which is the exact reason it becomes yours. It even arrives dressed as a compliment. “So organised.” “Such a calming presence.” “Honestly, great with people.” Every one of those is a genuinely nice thing to hear and a word-for-word translation of “will keep doing the unpaid coordination forever, and thank us for the trust.”

And the composure trap, the one where you are forbidden from showing the strain, is bolted a good deal tighter shut if you are a woman, because the readings are not symmetrical.

A man who lets a little of the pressure show is under pressure, carrying a lot, passionate about the work. A woman who shows the identical strain is not coping, is being a bit emotional, is perhaps not quite ready for the bigger role after all. So she holds it in longer and harder, with less permission to crack than anyone around her, and gets read in the end as either a machine or a mess, with almost no human band left in between.

The calm is not a temperament. It is a tax, it falls a great deal more heavily on some of us than others, and nobody ever sends an itemised bill.

The scream

The part nobody warns you about is the particular loneliness of being the stable one, which is that you are the single person in the building not permitted the reaction everyone around you is having freely and at length. Everyone else gets to be scared out loud. Everyone else gets to vent, spiral, catastrophise, book the emergency chat to be talked down off the ledge. You are the one doing the talking-down, which means your own fear has precisely nowhere to go, because the instant it shows, you stop being the fixed point, and the room does the arithmetic and decides that if even she looks worried, it is genuinely time to panic.

So you hold a completely straight face over a completely internal scream. You do it for weeks. You get unnervingly good at it, which is its own category of bad news, because a skill that useful to everyone else is a skill they will make very sure you keep supplying. I was, across that stretch, the calmest I have ever looked and the most quietly knackered I have ever been, and the gap between those two states is a debt. It always comes due later, and never anywhere convenient.

What I wish someone had said

I do not have a tidy five-step fix for this, and I distrust anyone who does, because the trap is built out of your own competence, and you cannot out-work a trap that runs on how much you can work. But a few things are true, and I wish someone had told me them while I was busy being load-bearing furniture.

The calm is not free, so stop handing it over at that price. The market rate for being the one person holding a company together through a crisis is not “thanks” and it is not “we really appreciate you.” It is money, or title, or scope, named out loud, in the moment, while the fire is lit and your value is impossible to argue with. Not at a review six months later when it is all calm again and everyone has gently forgotten who made it calm. Your leverage is never, ever higher than during the fire. Spend it during the fire.

Let a controlled amount of the strain show, upward, on purpose. Not a collapse of your own. A precise, deliberate sentence delivered to the person above the person who came apart: this is not sustainable, here is what breaks and roughly when, and I need either more help or a different plan. Total composure gets you handed more, because total composure reads as spare capacity. A visible, managed edge reads as a system nearing its limit, and a system nearing its limit is the only thing that ever persuades an organisation to shift a single ounce of weight off you.

Refuse to be the only adult. It landed on you partly by structure and partly because you let it, because letting it felt responsible and the alternative felt like abandoning people mid-drowning. It is not abandonment to say I can hold two of these, not five, and you will need to find owners for the other three. The company that quietly made five things your problem will just as quietly let them stay your problem forever, because from where it is sitting the problem is already solved. It is solved by you. At your expense. In silence.

The close

The deadline came. Something shipped. It was not the thing that had been promised, but it was close enough to a thing that everyone agreed, with audible relief, to call it done and never mention again. The infighting settled back into the low ambient resentment that passes, in most companies, for peace. The loud one, as far as I know, is still out there somewhere, explaining at volume that none of it was ever his fault. The founder learned sod all, because a founder committed to pleasing everyone is on a journey that only ends when they finally run out of everyone. And I was thanked, warmly and briefly, after which the calm I had personally manufactured quietly became the new baseline against which my next impossible request would be measured.

That is the real ending. Being able to hold it together is not a gift you have been given. It is a job description no one wrote down and no one is paying for, and the better you are at it, the more of it will be handed to you, forever, until the day you decide the scream is information rather than weakness and let it out somewhere it might actually do you some good.

I held it together. I would like it noted, for the record, that I did not enjoy a single second of it, and that the person everyone can rely on is, far more often than anyone will ever admit, simply the person who has not yet been given permission to stop.

I have, more than once, made a piece of work worse by refusing to stop working on it.

Not dramatically worse. Worse in the quiet way that only the maker can see, which is the most maddening way for a thing to be worse, because you cannot even claim the credit for the damage. The sentence that was perfectly clear at hour three, buffed into something cleverer and fractionally more opaque by hour seven. The design that was clean right up until I added the third refinement nobody had asked for, and one person gently asked me to remove. The deck that peaked, demonstrably, about four versions before the one I actually stood up and presented. I have a real talent for taking a finished thing and continuing to touch it until it is finished in a worse way, and I have donated to this practice, cumulatively, a genuinely appalling number of hours, every single one of them under the sincere and cheerful belief that I was helping.

I was not helping. Past a certain point I was doing something else entirely, wearing “improving it” as a disguise, and this article is about what that something else actually is, and how to make it stop.

Subscribe now

Perfectionism is not high standards

Let me clear one thing out of the way first, because otherwise the perfectionist will grab it and use it as a fire exit.

Perfectionism is not the same thing as having high standards. They are, most of the time, opposites wearing the same coat.

High standards are about the work. Perfectionism is about you. High standards ask whether the thing is good enough to do its job. Perfectionism asks whether the thing will expose you as a fraud, and then keeps polishing, not until the work is good, but until the fear goes quiet, which, being fear, it manages approximately never.

Those extra hours you pour into agonising over the final five percent are not, on the whole, buying quality. The quality clocked off a while ago and went home. What the hours are actually buying is a delay, a stay of execution before the work leaves your hands and becomes a thing that other people are allowed to have opinions about. That is the con at the centre of the whole habit. While the work is still “in progress” it cannot be found wanting, and neither can you, so the polishing continues, not as craft but as a rather elaborate method of never having to find out. You cannot fix a habit you keep flattering as diligence.

The opposite error is not the cure

Round about here, someone always produces the famous slogan about moving fast and breaking things, and offers it up as the antidote, as though the remedy for over-polishing were simply to stop caring and ship whatever falls out of your hands with the least resistance.

It is not, and I am going to be unusually firm about this, because that slogan has done nearly as much damage to good work as perfectionism ever managed, and it did it while looking cool, which is worse. Moving fast and breaking things is not a philosophy of quality. It is a philosophy of quietly relocating the cost of your mistakes onto other people, your users, your colleagues, the unfortunate soul holding the pager at 2am, and then describing the arrangement as boldness. Shipping something half-built at speed is not the opposite of perfectionism. It is the identical illness with the symptoms reversed and a better wardrobe.

Because the perfectionist and the move-fast merchant are, underneath, making precisely the same mistake. Neither of them has calibrated anything. One pours ten hours into a job that wanted two. The other flings two hours at a job that needed ten. They are both confidently wrong, in opposite directions, about one single question, and answering that question correctly is the entire skill this article has to offer. The rest is just me being rude about the two ways of getting it wrong.

The curve, and the knee

Every piece of work sits on a curve. Effort along the bottom, quality up the side. And the curve is not a straight line.

At the start, effort buys quality almost embarrassingly cheaply. The first pass drags the work from nothing to roughly right, and the returns are enormous. Then the curve begins to bend. Somewhere in the middle sits a knee, a point past which each additional hour buys a little less, and then less than that, until eventually you are spending a full hour of your one finite life to improve the work by an amount detectable only to you and, in time, possibly a forensic archaeologist.

The whole skill of knowing when to stop is the skill of finding that knee. Everything after this is decoration, some of it applied well past its own knee.

Stop before the knee and yes, you have genuinely under-baked it. Quality was still climbing, steeply and cheaply, and you walked away from returns that were sitting right there for the taking. That is the move-fast error, and it is a real one, whatever the mug on the desk insists. Grind on well past the knee and you are out on the long flat of the curve, paying the full hourly rate for improvements the world will never once notice, buying nothing at all except a few minutes of your own relief. That is the perfectionist error, and it is where I have personally parked most of a decade.

Good is not a sad little compromise you settle for. Good is the top of the curve. The knee is not the grubby midpoint between quality and speed where cowards meet for coffee. It is the point of maximum sense, and the perfectionist and the mover-fast are both standing a fair distance from it, each loudly insisting their distance is a personality.

Where the knee sits

The catch, and the reason this is a skill rather than a rule I could print on a mug of my own, is that the knee sits in a different place for every piece of work, and finding it is the actual job.

Price the stakes before you start, not after. The knee on a throwaway internal message sits at about four minutes, possibly three. The knee on a document a hundred people will read and later quote back at you sits considerably later. The knee on something you cannot take back, the migration with no rollback, the number you are putting in writing to someone who matters, sits later still. Perfectionism treats all three as though they were the migration. The cure begins with asking, before you have touched anything, what this specific thing is genuinely worth, and then flatly refusing to spend a product launch’s worth of care on a Slack message.

Sort the reversible from the permanent. Most work is a door you can walk back through: if it is wrong, you fix it, at a cost of mild and entirely survivable embarrassment. Only a few things are doors that shut for good once you are through them. A startling amount of perfectionism is simply standing guard over reversible work, defending it against a permanence it was never going to have. Ship the thing you can undo, and undo it if you must. Save the white-knuckle care for the door that actually locks. (This idea did a full shift in an earlier piece of mine on ambiguity, so forgive it for clocking on again here, but it turns out to be the same insight running in both directions, one for starting and one for stopping.)

Decide what “done” means before you begin. A task with no definition of done has no stopping rule, and a task with no stopping rule will cheerfully absorb every hour you are foolish enough to feed it, because there is always, always one more thing. So write it down first, in cold blood, before the fear gets a vote: what must this do, for whom, to what standard. Then when you reach it, you are allowed to stop, not because you have gone soft, but because a calmer and better-rested version of you already signed the permission slip.

Give it to a real human early. Your private sense of whether the work is good is almost always wrong, and wrong in the same direction, upward, because you are marking your own homework against an imaginary examiner considerably crueller than any actual person has ever been. A real reader relocates the knee for you, accurately, in roughly five minutes, at no charge. Hand the ugly draft to one trusted person, and their reaction will tell you what another solitary hour never could: whether you are honestly below the knee and not finished, or above it and simply frightened.

Watch who the last changes are for. The cleanest tell there is, so if you keep only one of these, keep this. Every time you catch yourself making a change, ask whether a single other human being would ever notice it. If the honest answer is no, then congratulations, you have stopped working on the piece. You are now working on your own anxiety, using the piece as a convenient surface, and the piece, for what it is worth, would rather like to be left alone.

The close

I would love to report that I have since learned to find the knee every time, and now release my work at the precise point of maximum sense, serene, untroubled, my evenings restored to me. I have not. The fear does not leave. I still catch myself at hour seven, lovingly sanding a perfectly good sentence down toward a worse one and calling it standards.

The only thing that has genuinely changed is that I now know what I am doing while I do it, which turns out to be most of the battle. The extra hours were never buying quality. They were buying a delay to the verdict, and the verdict is the only thing that ever tells you whether the work was any good, which means all that devoted polishing was not merely wasted. It was the precise thing stopping me from finding out.

You do not beat perfectionism by caring less. That is the move-fast lie, and it is a con dressed up as liberation. You beat it by caring about the right thing, which is not the work being flawless but the work being judged correctly for what it actually needed to be, and then finding the nerve to hand it over and let it be judged. Done is not the absence of standards. It is the highest standard going: knowing exactly what this particular piece of work was worth, and then declining, with some visible effort, to pay a penny more.

The last ten hours were never diligence. They were an invoice you keep quietly writing to yourself, for a service the work stopped needing hours ago, to put off a moment you were always, in the end, going to have to face.

I once sat in a performance review where I was, by every measure printed on the form, doing fine, and was made to feel I had let the entire species down.

The work was good. The projects had shipped. The little boxes were ticked in the little way that little boxes like. And yet the whole conversation had the texture of a gentle intervention, because I had committed the one genuinely unforgivable sin of the modern workplace: I appeared content. I was not visibly straining toward the next level. I had not, anywhere in my self-assessment, described my year as a “journey.” And when I was asked where I saw myself in five years, I left a pause slightly too long to be ambitious, because the honest answer was “roughly here, ideally with fewer meetings,” and I watched that answer land like a small and baffling crime.

I would like to talk about that crime, because I suspect a great many of us are quietly committing it, and then lying about it on our self-reviews in the approved dialect of hunger.

Subscribe now

The treadmill nobody agreed to get on

Somewhere in the last couple of decades, a job stopped being a thing you did and became a thing you were morally required to be ascending. The expectation now is not that you do your work well and go home to your life. It is that you are perpetually striving, growing, levelling up, leaning in, chasing the next title, optimising the self, building the personal brand, and doing all of it with the tireless enthusiasm of a golden retriever who has just been introduced to the concept of Q3.

Standing still is framed as failure. Being satisfied is failure with better lighting. There is now an entire content economy dedicated to informing you, daily, that if you are not becoming a slightly more impressive version of yourself at all times, ideally before 6am and ideally on camera, you are wasting the one wild and precious life that you were, apparently, issued a KPI for. Your hobbies should be monetised. Your rest should be “active recovery.” Your weekend should have a deliverable.

And I am, to use the industry term, over it.

Two lies, one con

The pressure arrives in two flavours, and they are the same con in two different hats.

The first is the passion lie: do what you love and you will never work a day in your life, find your why, follow your bliss cheerfully into an early grave. The second is the striving lie: you must always be climbing, because a person who is not climbing is, by definition, sliding, and probably being spoken about in a calibration meeting as “coasting.” Both lies do precisely the same job. Both are engineered to convince you that your job is not a job. It is a calling, or a ladder, or a journey of radical self-actualisation, anything at all except the mundane thing it actually is, which is an arrangement where you swap your finite and non-refundable hours for money.

The reason both stories are pushed so hard is not a mystery, and it is not a conspiracy either. It is just incentives. A workforce that believes its job is its passion will accept less money for it, because you are being paid partly in meaning, which does not show up on the payslip and cannot be spent. And a workforce that believes standing still is death will run up the ladder for free, on evenings and weekends and the good years of its one life, too frightened to stop. Passion keeps you cheap. Striving keeps you sprinting. Both are marvellously convenient for everyone who is not you.

The reframe, which is the only one you are getting

So. A job is a transaction. You give it competence and time. It gives you money. That is the deal, in full, with no hidden clauses about your soul.

It is a clean, honourable, entirely sufficient deal, and it curdles into misery only when you demand that it also provide your identity, your friends, your sense of purpose, your reason to exist, and a third-act redemption arc, and then feel personally betrayed when it turns out to be, this whole time, a job. The transaction was never the problem. The problem is being sold a romance and handed an invoice, and being told the disappointment is a failing in you.

You are allowed to simply do it. Well. For money. And then leave the building and have a life that the building knows nothing about.

How to run a job you are pleasantly indifferent to

None of this is an argument for doing your job badly, which would be its own kind of trap. It is an argument for doing it deliberately, so that it funds your life instead of quietly eating it.

Name what the money is actually for. “It’s just a job” is vague enough to feel bleak. Make it specific and it becomes a choice. The money is buying a mortgage, a runway, the freedom to leave later, the small daily fact of not being afraid of your bank app. A job whose purpose you have named is a decision. A job whose purpose you have not named is a room with no visible walls, which is how people end up living in them.

Keep the centre of your life somewhere the job cannot reach. The most efficient way to be destroyed by a job you feel neutral about is to let it become the whole of who you are, so that a bad quarter becomes an existential event. Put the good stuff, the load-bearing stuff, somewhere the org chart does not appear: a craft, a sport, people, a thing you build on the side purely because you like it and have no intention of putting it on LinkedIn. This is also, not coincidentally, the single best protection against burnout, because you cannot be hollowed out by a job that was never holding you up.

Do the work well, but for you, not for them. This is not loyalty. Competence is portable, and it is the thing that buys you the option to leave, so treat it as an investment in yourself rather than a gift to the shareholders. There is also a quiet, unfashionable dignity in doing a thing properly regardless of how you feel about it, which pays out in self-respect at a much better rate than the resentful eighty percent that a disliked job tends to invite.

Ration the discretionary effort like it is a finite resource, because it is. The people who burn out worst in jobs they feel lukewarm about are almost always the over-givers, pouring unpaid evenings and their full emotional bandwidth into something that is not going to love them back and cannot, being a job. Give it what it is paying for. Keep the surplus for the parts of your life that actually return the investment. This is not slacking. It is correct allocation, which is supposedly a skill we admire.

Put a review date on it. “Just for the money” is healthy as a season and quietly corrosive as an unexamined decade. The difference between the two is entirely whether you are still choosing it. So book the check-in, every six months or so, and ask the plain question: is this still funding the life, or has it slowly become the life. As long as the answer keeps coming back as a deliberate yes, you are fine. The danger was never the boring job. It was the boring job you stopped choosing and simply defaulted into while apologising to no one but yourself.

The one part I am not being flippant about

There is a difference between a job you do not love and a job that is harming you, and the cheerful indifference this whole piece is recommending is only for the first one.

A job that is merely unfulfilling is sustainable, and everything above is how to sustain it with your dignity intact. A job that is steadily eroding your health, your self-respect, or your grip on who you are is a different category of problem, and no amount of clever reframing will fix it, because the problem is not your attitude, it is the job. “I need the money” is true and reasonable, and it is also the exact sentence people use to talk themselves into staying somewhere genuinely damaging for years longer than they should. If that is the situation you are actually in, this piece is not for you, and the kindest thing I can say is that the cost of staying is usually higher, and much better hidden, than the cost of going.

The close

I never did learn to love that job, and I am not going to pretend, in the last paragraph, that striving was secretly the answer all along, or that indifference is its own kind of passion if you squint.

What I learned instead was smaller and far more useful. The job was fine. The money was a perfectly good reason. And the only genuinely broken thing in the arrangement was the ambient insistence that I should have wanted more, sooner, louder, forever.

You do not have to love it. You do not have to be climbing it. You are allowed to do good work, take the money, and be quietly, unremarkably, gloriously fine, which turns out to be the one thing the whole striving-industrial-complex is structurally unable to sell you, because there is no upsell on enough.

You are forty minutes into the final interview for a job you have decided you want. The hiring manager joined eight minutes late, apologised for being slammed, and has answered two messages since, thumbs moving below the desk in the belief that desks are opaque. You ask why the role is open. There is a small pause, a sentence that has clearly been said before, the previous person moved on, and then, slightly too quickly, a question about your notice period.

None of this will make it into your decision. You will file the lateness under busy, the thumbs under modern life, the pause under nothing at all, because you have decided you want the job, and a person who wants the job becomes remarkably talented at explaining away the smoke coming out of it.

Consider me the friend who grabs your arm and says: look again. Nine tells, in the order you are likely to meet them.

Subscribe now

1. The evergreen vacancy

The role has been open for six months, or it keeps reappearing with a lightly reworded description, or a recruiter contacted you about it in the spring and here it is again in the autumn. Roles stay open that long for a short list of reasons: the hiring bar is confused, the comp is under market and nobody will fix it, or people keep getting to the final stage, hearing something, and withdrawing. All three are the same underlying condition. The organisation cannot agree what it wants or will not pay for it, and the vacancy is where that argument is being stored.

2. The missing veterans

Count the tenure in your interview loop. If everyone you meet joined within the last year, you are not looking at a growing team. You are looking at a replaced one. Growth adds people around a stable core; fire replaces the core. Ask each interviewer how long they have been there. It’s the single highest-yield question available to you, and it doesn’t even sound like a question.

3. The missing win

Ask what the team shipped recently that they are proud of. A healthy team answers instantly and specifically, and usually argues with itself about which thing to name. A burning team reaches for a process achievement (”we’ve really matured our planning”), a future achievement (”the next quarter is when it comes together”), or a diplomatic abstraction (”so many things”). Teams that cannot name a win have not had one recently. Teams that have not had one recently are under exactly the kind of pressure that lands on the newest person.

4. The unexplained backfill

“Why is this role open?” is a normal question with a normal answer. Someone was promoted, someone left for a bigger job, headcount grew. When the answer is a small verbal shuffle (”the previous person moved on”, said quickly, with a segue), you have found something. People do not become vague by accident in interviews; vagueness is effort. The follow-up is polite and lethal: “What would the previous person say was the hardest part of the job?” Watch the weather change.

5. The family

“We’re like a family here” is offered as warmth and functions as a warning. Families do not have performance reviews, exit processes, or on-call rotas. What the phrase reliably predicts is the absence of boundaries: the expectation that you will absorb unreasonable things out of loyalty, and the framing of any refusal as a betrayal rather than a negotiation. Healthy teams describe themselves in terms of what they do. It is the burning ones that need you to love them. Run.

6. The process is the sample

Interviews rescheduled twice. An interviewer who clearly hasn’t read your CV. Two people in the loop who describe the role differently. A take-home task that nobody gives you feedback on. Candidates persistently forgive all of this as admin noise, and it is not admin noise. Hiring is the process the organisation runs when it is trying its hardest. If they can’t coordinate five calendar invites while courting you, extrapolate honestly to the quarter when they are not.

7. The proud exhaustion

Listen for how workload is described. “It’s intense, but we’re intense people.” “We move fast.” “No two days are the same.” Fatigue, worn as identity, is the most reliable single indicator on this list, because the people saying these sentences believe they are selling you something. Bonus data if an interviewer answers messages during your interview, apologises for being “slammed”, or takes the call from a train. That is not a busy week. That is a recipe for burnout.

8. The undefined win condition

Ask the hiring manager what success looks like in this role at six months. A functioning team has thought about this, because the role exists to fix or build something specific. The burning-team answer is a shape-shifter: “hit the ground running”, “we’ll figure that out together”, “honestly, just taking things off my plate”. A role with no definition of success is a role that cannot be succeeded at, and it usually means you are being hired as capacity for a fire, not as the owner of anything. Capacity for a fire is consumed by it.

9. The exploding offer

The process crawled for weeks, and now suddenly the offer expires on Friday and everyone is calling you. Manufactured urgency at the close is not enthusiasm. It’s the same move a timeshare uses, and for the same reason: the product survives excitement better than it survives inspection. A team confident in what it is offering gives you time to inspect it. The pressure to decide quickly is, always, information about what a slower look would have found.

What to do with a positive reading

One tell is weather. Every team reschedules occasionally, and some healthy teams are genuinely intense in bursts. The signal is in accumulation: two of these is a conversation with yourself, three or more is a pattern, and a pattern at best-behaviour stage is a promise about ordinary Tuesdays.

None of this necessarily means walk away. Sometimes the burning team is the promotion, the mission, or the money, and you take it with open eyes, priced in, on your terms, which mostly means: in writing, with a scope, and at a number that reflects what you now know. What the tells are for is not avoiding every fire. It is making sure that if you walk into one, it was a decision, and not a discovery.

The team on fire will not tell you it is on fire. But it cannot stop the smoke coming out in the interview, because the interview is made of the same people, under the same conditions, doing the same things. They were always going to show you. You just have to stop marking it down as nerves, admin, or passion.

I had one meeting. I did not understand the weight of that at the time.

My manager was three months into a sick leave that had quietly deleted him from every process that mattered, and in the vacuum I had been left, in the DeepMind way, to simply get on with it. So when I finally got half an hour with my skip-level, a man too busy to have found the time before I asked for it directly, I came prepared. I laid out everything I had already done. Not planned to do. Done. Shipped, owned, closed. I had been running a level above my title for the better part of a year, I had the receipts, and I walked him through them.

He listened. And at the end he said, in the offhand way people say the things that will keep you awake for a fortnight:

“Yeah, that sounds like L4 work.”

It felt like taking a breath I had been holding for a year. After all those months of doing the work in silence, of treating my own invisibility as a discipline, one sentence from the one person who mattered, and I had, at last, been noticed for all of it, and was, I let myself believe, rightly about to be rewarded for it. The relief was almost physical. For about ten days I believed the system had a sensor after all, and that it had finally registered me.

The email came two weeks later.

He did not think I was ready. I was welcome to write a promotion packet if I wanted to, but in the interest of being honest with me, it was likely to be rejected.

Read that twice, because I did, many times, hunting for the version of it that reconciled with the meeting. There isn’t one. There is no reading in which “that sounds like L4 work” and “I don’t think you’re ready and you’ll probably be rejected” are describing the same body of evidence. Nothing had changed in two weeks. I had not un-shipped anything. The only thing that had happened between the sentence and the email was time, and whatever was said, or not said, in rooms I was not in.

What I felt, in order, was confusion, then a specific hot flush of anger, then something quieter and worse, which was the sense of having done a great deal of work for nothing, of being invisible in a way that no quantity of output was going to fix. I had brought the receipts. The receipts were not the currency. I had turned up to a game believing it was scored on the thing I was good at, and found out halfway through that it was scored on something else entirely, something nobody had named, and that I held no chips for.

There is a particular cruelty in the construction of that email that took me a while to figure out. “You can write the packet, but it will probably be rejected” is not guidance. It is a manager moving risk off himself and onto you. He is not telling you the odds to help you. He is building the record in which, however it goes, he warned you, he was reasonable, he was honest, and the failure was yours for proceeding. The pre-emptive no is a way of being on record as not having backed you while appearing to leave the door open. It is deniability, dressed as candour. And it works, because most people, correctly reading the signal, do not write the packet at all.

I wrote the packet anyway.

I would like to say I did it out of defiance, but it was closer to disbelief, a need to make the system reject the actual evidence rather than a vibe in an email. So I assembled the whole thing, a year of operating a level up, and I submitted it into the machine, and the machine did exactly what he had told me it would do.

It rejected me.

And then, a cycle later, with no meaningful change in what I was doing, no new scope, no grander project, nothing I could point at and say there, that is the thing that tipped it, they promoted me.

Sit with that for a second.

If the work was L4 the second time, it was L4 the first time. The work was identical. Which means “not ready” was never a description of me, or of my output, because my output did not move. It was a description of something else, something the organisation was too polite or too unaware to say out loud, and the promotion a cycle later is the proof. It is the control experiment I never designed but got anyway. You change one variable and hold the rest constant. The variable was not the work. The work was constant. The variable was time, and what accrued in that time, which was not competence. It was position.

Turns out, a promotion is not a measurement. It is a story the organisation tells about you, in a room you are not in, and the story is authored by your visibility and your allies, not by your work. Your work is the raw material. The story is the product. “Not ready” is what the machine returns when the story has not yet been written by enough of the right people, and it will happily hang that label on a body of work it later promotes unchanged, because the label was never about the work. It was about the story, and the story had simply not been told yet.

I told myself, for a while, that the people who got promoted ahead of me on thinner work were better in some way I couldn’t see. They were not. I watched them closely, because I was trying to learn the trick. They were louder about what they did, and they had built better alliances with the people who spoke in the rooms. That was the whole difference. The same work, sometimes less of it, described more often, to more of the right people, by more voices than just their own.

This is where it stops being a general story and becomes the specific one I write this column about, because loudness and alliance are not neutral tools equally available to everyone. Being visibly, repeatedly vocal about your own achievements reads as confidence in some people and as arrogance in others, and which one it reads as has very little to do with the achievements and a great deal to do with who is doing the talking. The people who could be loud without a penalty were, broadly, the ones the culture already read as leadership-shaped. The rest of us were handed the advice that sounds like fairness and operates as a trap: keep your head down and do great work, let the work speak, it will be recognised.

The work does not speak. I have said it before and I will keep saying it, because it is the load-bearing lie underneath all of this. The work is mute. It is spoken for, or it is not, and the entire apparatus that decides who gets to speak for it is exactly the apparatus that “do great work and stay humble” trains you to ignore. The meritocracy is not a description of how the place runs. It is a tax, and it is levied hardest on the people naive enough to believe in it, who are disproportionately the people who were never going to be handed the benefit of the doubt in the first place.

So, the things I would tell the version of me sitting stunned in front of that email. They are more cynical than anything I would put in front of the paywall, because they are the actual mechanics rather than the respectable account of them.

Read more

I don’t know about you, but I’m tired.

It is now July (which I am fairly sure it was not a moment ago), and the year has the particular quality of having both dragged and vanished at the same time. Everything has felt relentless lately: the work, the news, the low background hum of an industry that keeps making people redundant and filing it under correction. I have spoken to a lot of people in tech over the past few months, and underneath whatever they were ostensibly talking about, most of them were saying a version of the same thing. I am so tired. And I do not feel allowed to be.

Let me say this: if you are tired right now, you are not failing at something. You are responding accurately to conditions that are genuinely hard.

For a long time, my own response to that tiredness was to refuse it. I operated on a private theory, never quite said aloud, that being relentlessly available was a form of redundancy insurance.

If I answered the late message, took the call on the day off, worked through the cold I should have stayed home with, never fully logged off on holiday, and was visibly, demonstrably more relentless than the people around me, then when the cuts came (and in this industry the cuts always come) I would be too obviously essential to be on the list. Relentlessness was the premium. Safety was the policy. I paid the premium for years, in evenings and weekends and a nervous system that had forgotten what its resting state felt like.

Then a round of cuts came through, and it took people who were every bit as relentless as I was. Some of them more so. The theory did not just fail to pay out. It turned out never to have been a real policy at all. There was no underwriter. I had been voluntarily paying a premium on insurance that did not exist, and the only thing the payments had reliably bought me was a body that arrived at the single most demanding moment of my career (the aftermath, the market, the scramble) already running on empty, with nothing left in the tank for the part that actually mattered.

This piece is about why that was a predictable engineering failure, and not a moral one. And about why, in a market this grim, the instinct almost everyone has, to grind harder and sprint at the redundancy, is exactly, mechanically wrong.

Subscribe now

The relentlessness is real, and so is the fear

Tech right now is genuinely relentless, and the fear is genuinely rational. The layoffs are real and they are everywhere. The job market is, by most honest accounts, the worst it has been in over a decade. Roles that would have taken three weeks to fill now take six months to find, and the queue for every opening is out of the door. In that environment, the instinct to clamp down and grind harder is not stupidity. It is a sensible-feeling response to a frightening situation. You can feel the logic of it: work is scarce, so cling to the work; safety is thin, so perform indispensability.

The logic feels right. It is also, on inspection, the thing that breaks you, and it breaks you at the worst possible time.

You cannot sprint a marathon, and the market just made the marathon longer

A sprint and a marathon are not the same activity at different lengths. They are different activities that happen to both involve running. Run a marathon at sprint pace and you do not finish slightly out of breath. You do not finish at all. You fold up at the side of the road somewhere around the eighth mile, and the people who paced themselves jog past you while you are being attended to. Trust me.

Everything about the current market makes this longer, not shorter. A bad job market does not compress the race; it extends it. The thing you are trying to outlast (the downturn, the hiring freeze, the months of applications) is a marathon by definition, and the people sprinting at it are guaranteeing they hit the wall long before the finish line comes into view. The grind that feels like effort toward safety is actually the fastest route to being depleted at precisely the moment you most need to be sharp: the interview, the pivot, the hard conversation, the decision about what comes next.

You do not protect yourself from a long, hard stretch by going flat out at the start of it. You protect yourself by still being upright at the end.

You are a system running at one hundred percent utilisation

You are a system. Systems have a sustainable operating load, and it is not one hundred percent. Run any system at full utilisation continuously and you do not get maximum output. You get a system with no headroom, and a system with no headroom does not degrade gracefully. It does not slow down a little when an unexpected load arrives. It throttles, it queues, and then at the first genuine spike it falls over completely, taking everything with it. The difference between a system running at seventy percent and one running at a hundred is not thirty percent more output. It is the entire margin you need to absorb the thing you did not plan for. And there is always a thing you did not plan for.

Rest is not the reward you get for running hot. Rest is the maintenance window. It is the scheduled downtime that exists so that you do not take unscheduled downtime, the kind that arrives without warning, lasts months instead of hours, and is not in your control. Every reliable system in the world has planned maintenance built in, and nobody running those systems thinks of the maintenance window as weakness or indulgence. They think of it as the reason the system is still up.

This is the actual answer to the line about self-care being selfish, which I am otherwise allergic to, because it has been said into the ground by people selling candles. It is not that rest is a virtuous thing you should feel permitted to do. It is that you do not congratulate a server for refusing to reboot until it melts. You do not call the deferred maintenance “dedication.” You call it the incident report.

Maintenance, for people who will try to optimise it

There is a specific trap for our kind of person. It is where most tech people’s attempts at rest go to die: we do not struggle to rest because we are lazy. We struggle because we turn rest into another project with metrics. The sleep score. The recovery percentage. The readiness number off the ring. The quantified self is just work wearing a tracksuit, and a rest you are grading yourself on is not rest. It is an unpaid second job with worse documentation.

Actual recovery has no KPI. With that as the only rule, a short list, framed as maintenance rather than optimisation:

Sleep. The activity, not the metric. Do the thing the ring measures instead of measuring it. This is the single highest-leverage maintenance task available to a human being and it cannot be hacked, only done.

Perform a hard shutdown. Take the work messaging off your phone (guilty). Latency is allowed. You are a person, not an on-call pager, and the belief that you must be reachable is a story you are telling yourself, not a clause in your contract.

Use your hands on something with no output anyone will see. Cook the slow thing. Fix the engine. Play the instrument you are genuinely bad at. The point is a task whose entire value is that it produces nothing legible, cannot be put on a profile, and resists being turned into a side hustle. I literally quit DeepMind and ended up building cars for 2 years. Working on something that existed in real life and not the ether healed my soul.

Move, with the tracking off if the tracking has become the point. A walk, a lift, a row, done as maintenance and not as another leaderboard. If you cannot do the exercise without competing at it, that is the tell, and the fix is to do it badly on purpose. But please, for the love of god, lift with your legs and not your back.

Allow boredom. Take the walk with no podcast, no audiobook, no input at all. A system needs idle cycles to run its background processes, and so, it turns out, do you. Ignore the discomfort you feel in the first ten minutes of unstimulated quiet. It’s worth it.

See people who do not work in tech and will not, at any point, ask you about your roadmap.

None of this is profound. Profundity is also a kind of effort, and the entire exercise is to stop, briefly, making everything an effort.

One honest line

There is a difference between tired and something heavier. If easing off for a weekend and sleeping properly puts you roughly back together, this piece is for you, and the engineering holds. If you ease off and the flatness is still there on Monday, and the Monday after that, then what you are dealing with is not a pacing problem and it will not yield to a better-managed stint. It is worth talking to someone who is not me about it, and worth taking at least as seriously as you would take a production system that stayed down after the reboot.

The close

I would love to tell you I have fixed this. I have not. There is a reasonable chance I am writing this later in the evening than I should be, having spent the day not taking my own advice, which is the standard hypocrisy of anyone who writes things like this, and you should assume it of me.

The point was never that I cracked it. The point is narrower and, I think, more useful: the instinct to grind through, to sprint at the danger, to treat relentlessness as insurance and rest as the thing you earn afterwards, that instinct is a bug, not a virtue. It feels like strength and it performs like a system with no headroom, which is to say it works right up until the exact moment you needed it to hold.

Take your foot off the gas. Not because you have earned it. Because the race is long, the road is worse than usual right now, and the only version of you that gets to the end of it is the one that did not redline the whole way there.

There is a specific genre of post you have all seen, in which someone announces, with the breathless joy of a person unveiling a cure for a disease, that they are Going Paid. There is usually a countdown. There is often a graphic. There is always a sentence that begins “I’m so excited to...” and ends somewhere near your wallet.

This is not going to be that. Partly because it is not in me, and partly because I recently wrote two thousand words arguing that work is an honest transaction, and it would be a bit much to abandon the principle the first time money is actually on the table.

So. The honest, slightly British, faintly embarrassed version.

I am adding a paid tier. Today. And the most important thing I can tell you about it is the part that is not changing.

Subscribe now

The essays stay free. All of them. Forever.

The Tuesday and Friday pieces are not going behind a paywall, now or later. The writing that brought you here is the writing that keeps you here, and locking it up the moment a few of you turned up would be both rude and strategically idiotic. If you never pay me a penny, you lose nothing. Not one word. I wanted that established before the part where I ask for money, because the asking is considerably less excruciating when nobody is being held to ransom.

Now, the asking.

The unglamorous truth is that writing this properly, twice a week, is a second job. It is a second job I do in the evenings, after the first job, occasionally while muttering. I am not on the breadline and I am not going to perform a hardship I do not have. But there is a difference between a thing you do for love and a thing you do for love at eleven at night for the second time that week, and the difference is roughly the price of a paid subscription.

If the free essays have earned a place in your week, paying is simply how you keep them coming, and how you keep them free for the people who genuinely can’t.

What you actually get for your money.

Three things, chosen on the strict principle that I had to genuinely want them myself or they did not make the list:

• Blockers. A weekly chat where you bring me a real problem from your role. Not the version you would put in a status update. The version you would only say in the pub. The decision that has been “circling back” since the spring. The colleague who is, whatever the org chart insists, the actual blocker. Think of it as me in your corner, mildly caffeinated, every week.

• The scripts and templates. The essays describe the manoeuvre. These hand you the actual words. The manage-up emails, the salary-standoff script, the pre-mortem agenda: copy, paste, send, get on with your day. The theory is free. The homework is paid.

• Paid-only Incident Reports. The stories I can only really tell with the door shut. Too specific, too candid, or too likely to be recognised to put on the open internet where the wrong person might read them over their lunch. These are the ones with the good bits left in.

On price, briefly.

It is the price of roughly one coffee a month, which feels like the correct order of magnitude for a thing that turns up twice a week and occasionally saves you from a bad meeting. There is no launch discount, no countdown, no founding tier, no tiered scheme of escalating perks. One price, one thing, and you can leave whenever you like.

No graphic, no countdown, no joy so excessive it requires medical attention. If these essays have been worth your time, this is me asking, with the appropriate quantity of British discomfort, whether you would help keep them going. And if not, the essays are free anyway, I am not going anywhere, and I will see you on Friday.

Subscribe now

The first time someone handed me genuine ambiguity, I treated it as a mistake.

I had been given a mandate that amounted to “own this area, figure out what needs doing, and do it.” There was no definition of done. There was no clear owner of the thing I was supposedly now owning. There was no success criteria, no spec, no list of requirements I could work through and tick off and hold up as evidence that I had finished. I assumed this was an oversight. So I did the diligent, conscientious, completely wrong thing: I went looking for the missing clarity. I asked for the spec. I asked who owned what. I asked what success looked like.

The answer, eventually delivered with a slightly puzzled expression, was: that’s what we’re hiring you to figure out.

The ambiguity was not a precursor to the job. The ambiguity was the job. And I had spent three weeks treating the central task as an administrative error I could escalate my way out of.

This is the thing nobody quite says plainly about “comfortable with ambiguity,” which appears in approximately every job description above a certain level and is almost never explained. It is not a personality trait. It is not a calm temperament or a high tolerance for chaos or the ability to look serene while everything is on fire. It is a specific, learnable, mechanical skill, and the phrase does it a disservice by making it sound like a mood.

Subscribe now

What the phrase is actually hiding

“Comfortable with ambiguity” sounds like it is describing how you feel. It isn’t. Nobody is genuinely comfortable not knowing what they are doing. The senior people who look comfortable are not experiencing less uncertainty than you, they have simply stopped requiring the uncertainty to go away before they act.

That is the whole skill, and it is worth stating in plain terms because the soft framing obscures it: the ability to operate in ambiguity is the ability to act on incomplete information, build your own provisional structure where none has been provided, and correct as better information arrives, instead of waiting for a certainty that is not coming.

The person who is “bad at ambiguity” is not weak or anxious or unsuited to senior work. They are usually just doing what they were trained to do for twenty years of education and early career: waiting to be given the spec, completing the spec, and being rewarded for completing the spec accurately. That training is precise and it is useful and it stops working the moment the spec stops arriving.

The reframe: there is no spec, and you are allowed to write it

Ambiguity is the absence of a spec. The instinct is to treat that absence as a blocker. To wait for someone to produce the missing document. But in genuinely ambiguous work, no one is going to produce it, because no one has it, because if the answer were known the work would not have been handed to a person at your level to figure out.

So you write the spec yourself. Provisionally. You make a set of assumptions about what the goal is, what done looks like, who the constraints belong to, and what you are going to do first. You write them down, mark them clearly as your current best guess, and act on them. Not because they are right. They are almost certainly partly wrong. But a wrong spec you can correct beats no spec you are paralysed in front of, every time, because the wrong spec generates something to react to, and reactions are information.

This is the inversion that takes people too long to see: in ambiguous work, you do not think your way to clarity and then act. You act in order to generate the clarity. Action is the instrument that produces the missing data. You cannot get it by staring harder at the fog.

The actual mechanics

A handful of moves that turn the abstract skill into something you can do on a Tuesday.

Separate what you don’t know from what you’re only afraid you don’t know. Ambiguity panic inflates. When you sit down and actually list the genuine unknowns versus the things you do in fact know but feel uncertain about, the real unknowns are usually a short, manageable list. Write both columns. The fog is mostly fear; the map is mostly already in your head.

Write the strawman and circulate it. Produce your provisional spec. “Here is what I’m assuming, here is what I think done looks like, here is what I’m going to do first”, and send it to the people around the work. This does two things at once. It creates structure where there was none, and it flushes out the real constraints and the real owner faster than any number of clarifying questions, because people who will not define a problem for you will absolutely correct a definition you put in front of them. A wrong strawman is bait. The corrections are the spec.

Make reversible decisions quickly and irreversible ones slowly. Most ambiguity tolerance is really decision-velocity calibration. The reason you are frozen is usually that you are treating every decision as irreversible and therefore high-stakes. Most are not. If a decision is a two-way door, if you can walk back through it cheaply, make it fast on incomplete information, and correct later. Reserve the slow, careful deliberation for the genuinely irreversible calls. Sorting your decisions into these two buckets is most of the job.

Ship something small to buy information. When you cannot see the path, take the smallest concrete action that will tell you something true, and take it now. A rough draft, a prototype, a single conversation, a one-page proposal. The point is not the artefact. The point is that putting something real into the world produces a reaction, and the reaction tells you where the edges of the problem actually are. You are probing, not finishing.

Give yourself a decision deadline. The failure mode of ambiguity is not bad decisions, it is the infinite loop of gathering more context to avoid making one. Set a date. “I will decide the direction by Thursday with whatever I know by Thursday.” A deadline you impose on yourself converts open-ended dread into a bounded problem, and bounded problems are survivable.

Know which kind of ambiguity you are in. Some ambiguity is resolvable. There is an answer, someone has it, and the right move is to find that person and ask the precise question. Some ambiguity is irreducible. No one knows, there is no answer yet, and the work is to decide. Spending days hunting for an answer that does not exist is as much a failure as failing to ask the one person who had it. Diagnose which you are in before you choose your response.

The thing nobody tells you

The senior people you watch operating calmly in chaos are not less afraid of the fog. I want to keep saying this because it is the part that would have saved me the most time. They have simply internalised that the fog is the normal weather of interesting work, that waiting for it to lift is the same as waiting forever, and that the way through is to move while you cannot see clearly and adjust as the shapes resolve.

You do not become comfortable with ambiguity by feeling comfortable. You become comfortable with it by acting competently while uncomfortable, enough times that the discomfort stops registering as a reason to stop. The feeling never fully goes. You just stop letting it hold the controls.

The spec is not coming. It was never coming. The job was always to write it yourself, in pencil, and start.

A few years ago I sat in a routine one-to-one and my manager asked, pleasantly, “so what are you working on at the moment?”

It is a harmless question. It is also the moment I understood, with a small cold lurch, that this person genuinely did not know. Not because they were negligent, they had eight reports and a job of their own, but because I had told them almost nothing. I had kept my head down, done the work, and assumed that the doing of it was self-evidently visible. It was not. I was sitting across from the one person whose entire job included representing me in rooms I would never enter, and I had handed them almost no data to do it with.

That manager was going to go into a calibration meeting in a few weeks and advocate for me, or fail to, on the basis of whatever they happened to know. And whatever they happened to know was whatever I had bothered to tell them, which was nearly nothing, which meant they would improvise. People improvise from defaults. The defaults are rarely in your favour.

Managing up is the practice of fixing this. It has a bad reputation because it gets confused with flattery, with politicking, with being the person who laughs too hard at the skip-level’s jokes. That is not managing up. That is just being tiring. Managing up is narrower and more boring and far more useful: it is controlling the inputs your manager runs on, so that when they represent you, and they will, constantly, in your absence, they do it with your version of events instead of a guess.

And the medium for almost all of it is writing. Specifically, writing your manager can forward. Because the sentence you write in an email is the sentence that gets pasted, unedited, into the calibration doc, repeated to the VP, dropped into the planning thread. Your manager is a distribution channel, and the scripts below are just you giving the channel something accurate to distribute.

Six of them. Steal whichever you need.

Subscribe now

1. The Receipt

Situation: Your manager cannot represent work they do not know about. Send the update before they ask, in language they can forward without rewriting.

Subject: [Project] — where it landed this week

Quick note so you’ve got it ahead of [planning / calibration / the leadership sync].

This week I shipped [specific thing], which means [outcome in their language, risk removed, time recovered, the fire that didn’t happen].

Next week I’m focused on [thing]. The one I’d flag as at risk is [thing], because [reason]. Shout if you want to weigh in.

Why it works: the line “[specific thing], which means [outcome]” is the sentence your manager copies straight into the document that decides things. You have written your own metadata instead of leaving it to be written by someone with less interest in getting it right.

2. The Flare

Situation: A blocker you swallow in silence becomes your fault the moment it lands. Raise it early, and name the ask, not the anxiety.

Subject: Blocked on [X] - need a decision by [date]

I’m blocked on [thing]. To clear it I need [the specific ask: a decision, an introduction, sign-off, an hour of someone’s time].

If I don’t have that by [date], the knock-on is [concrete consequence, with its own date].

My recommendation is [your proposed fix]. Happy to go another way if you’d prefer.

Why it works: it converts a vague worry into a dated, owned decision with a default attached. Managers act on asks, not on ambient dread. And it is on the record, so the timeline cannot later be rewritten into a story where you sat on it.

3. The Reconciliation

Situation: You have been handed more than fits. Do not silently absorb it, and do not silently drop one and hope nobody notices. Push the trade-off back up to where it belongs.

I want to make sure I’m pointed at the right things. I’m currently holding [A], [B] and [C], and I don’t think all three land at the quality we’d want on this timeline.

If I’m protecting one, I’d protect [A], because [reason]. Which of the others would you rather I let slip, or take off me entirely?

Why it works: it refuses the trap where you either burn out trying to hit all three or quietly fail one and own the failure alone. Prioritisation is your manager’s decision. This hands it back to them in plain sight, with your recommendation already attached, so the default outcome is the one you wanted.

4. The Pre-Read

Situation: Your work is about to be discussed in a room you are not in. Give your manager the accurate version before someone hands them an inaccurate one.

Subject: For [meeting] - quick arming note on [topic]

Before [meeting], the short version so you’re covered:

• What it is: [one line]

• Where it stands: [one line]

• The thing likely to come up: [the hard question, and the answer]

• What I’d want represented: [the single sentence you want said on your behalf]

Why it works: you are writing the sentence that travels into a room you cannot enter. Supply it, and your manager carries your version. Don’t, and they improvise under time pressure, which defaults to whatever is easiest to say, which is almost never the version that serves you.

5. The Flag

Situation: You think a decision is wrong. Saying so only in your head helps nobody. Saying so only in the corridor afterwards helps nobody and costs you your credibility. Put it in writing, once, cleanly, then let it go.

I want to flag a concern on [decision], on the record, and then I’ll commit either way.

My worry is [specific risk, and why]. If we’re taking that on knowingly, I’m genuinely fine to proceed. I’d just rather it be visible now than discovered later. What’s the thinking?

Why it works: “on the record, and then I’ll commit either way” is the phrase that lets you disagree without becoming the person who relitigates a closed call. You register the dissent, you don’t block the team, and if it does go the way you feared, the record shows the risk was named, which is the entire difference between a team that learns and a team that goes looking for someone to blame.

6. The Runway

Situation: Your manager is not a mind reader. If you want a promotion, a different kind of work, or more scope, they cannot build toward it while they don’t know it exists.

I’d like to be working towards [staff / lead / owning a domain] over the next [timeframe].

I’m not asking for it today. I’m asking what the gap looks like from where you sit, and which work over the next couple of quarters would close it fastest. I’d rather aim at the real target than guess at it.

Why it works: it turns a private hope into a tracked objective your manager is now implicated in. Ambition you have stated is something they can sponsor. Ambition you have kept to yourself is something they will quietly assume you don’t have, and the career-making work gets allocated to whoever did say it out loud.

The thing underneath all six

None of these is flattery. None of them asks you to laugh at a joke that wasn’t funny or to spend your evenings cultivating anyone. They are all the same small, unglamorous move: making sure the accurate version of events reaches your manager before an inaccurate one does, in a form they can pass on without effort.

Your manager is going to represent you whether or not you help them. They are going to describe your work, weigh your priorities, defend your timeline and speak your name in rooms you will never see, using whatever inputs they happen to have. The only question these scripts answer is whether those inputs are yours, or a guess assembled in your absence.

Head down and hope is also a strategy. It is just the one where someone else writes your inputs for you, and you find out what they wrote at the same time everyone else does.

Some years ago I delivered the hardest piece of work I had done up to that point. It was the kind of thing that does not go well and went well anyway, mostly because I refused to let it not.

A few weeks later it was presented to leadership. Not by me. By someone who had been adjacent to it. Close enough to describe it, far enough to describe it wrong. In the readout it became a team effort with no particular author, delivered “with support from” a list of names on which mine appeared fourth. The framing was not malicious. It was just the framing of a person describing work they had not done, reaching for the nearest available narrative, which happened to be the one that served them.

I was in the room. I had a clear, factual, completely defensible correction available to me. The work was mine and I could have said so. And I sat there and said nothing, because saying something felt grubby, self-promoting, political, beneath me, and because I believed, with the quiet confidence of someone who has not yet been through a promotion calibration, that the work would speak for itself.

It did not speak for itself. Six months later, at calibration, the work spoke in the voice of the person who had described it in the room. It was weighted as a minor contribution to someone else’s win. I had done the work. He had done the distribution. The system paid out on distribution.

This piece is about why I was wrong, and about the specific, recoverable mistake underneath it, which is not a mistake of character but a mistake of architecture.

Subscribe now

“I just focus on the work” is a strategy.

Here is the sentence, in some form, that nearly every good engineer and operator I know has said with pride at some point: I don’t do politics. I just focus on the work.

I said it for years. The appeal is real, so I want to be fair to it. It feels like integrity. It feels like a refusal to debase yourself in the grubby status games that lesser people play. It feels like keeping your hands clean.

It is a strategy. It is just almost never the strategy the person saying it thinks it is.

What “I just focus on the work” actually does is delegate every decision about how your work is described, attributed, surfaced and weighted to other people. It does not remove you from the political system. It removes your input from the political system while leaving you fully subject to its outputs. You have not opted out. You have opted out of the part where you get a say.

Heads-down is not a neutral position. It is a choice, usually an invisible one, to let the room write your story in your absence. And the room will. The room always does. Nature abhors an unattributed piece of work.

Work is the product. Politics is the distribution.

Here is the frame that took me an embarrassingly long time to see, put in the only terms I find convincing.

The work you do is a product. It can be excellent. It can be the best product in its category. But you already know, because you have watched it happen to better products than yours, that the best product does not win. The product with the best distribution wins. The one that gets in front of the right people, described in the right language, at the right time, attached to the right narrative. Distribution beats quality, not because quality does not matter, but because quality with no distribution never reaches the people who decide what matters.

Office politics is distribution.

That is the whole reframe. Politics is not a corruption sitting on top of a clean meritocratic signal. There is no clean signal underneath. Politics is the layer through which the organisation transmits information about who did what and what it was worth. It is the routing protocol. It is how the payload of your work actually moves through the system to the people with the budget, the headcount and the calibration spreadsheet.

When you refuse to engage with that layer, you are not keeping your product pure. You are building something excellent, shipping it to a system with no distribution attached, and then watching, puzzled, and slightly wounded, as a worse product with a louder release note gets the install base.

The engineer who says “I just focus on the work” has built a great product and declined to ship it anywhere. And because the work still has to be described by someone, in some language, in some room, the description gets written by whoever is willing to write it. Usually that is someone with more to gain from describing it a particular way than you had to gain from describing it accurately.

The thing the apolitical engineer gets backwards

The deepest version of the mistake is a belief about fairness. The apolitical engineer believes, has been raised, schooled and onboarded to believe, that there exists, somewhere above them, a fair evaluator. A manager, a committee, a process, a system that, left alone and supplied with good work, will eventually see clearly and reward correctly.

There is no such evaluator. There is only the system, and the system does not see. It reads. It reads whatever was written into it, the readout, the brag doc, the calibration packet, the Slack thread the VP happened to scroll, the sentence your skip-level repeats because it was the sentence they were handed. If you wrote nothing into the system, the system did not read nothing. It read what other people wrote about you.

The absence of your input is not an absence in the record. It is just someone else’s input, uncontested.

This is the part that should reorganise how you think about it: opting out of politics does not produce a fairer outcome for you. It produces someone else’s preferred outcome, frictionlessly, because you removed the only person in the building with both the accurate information and the incentive to enter it.

What to actually do, without becoming the person you don’t want to be

The objection here is always the same, and it was mine too: I don’t want to become one of those people who spends more time managing their image than doing the job.

Good. You don’t have to. There is a precise line between self-promotion and distribution, and the whole game is staying on the right side of it.

Self-promotion is claiming work you did not do, or inflating work you did. Distribution is making sure the work you actually did is accurately described, by you, before it is inaccurately described by someone else. The first is grubby. The second is hygiene. The apolitical engineer collapses the two into a single forbidden category, refuses the whole category, and that is how they end up fourth on a list of names.

Three moves, in increasing order of discomfort.

Write the description before someone else does. The highest-leverage political act available to you is also the most boring: write down what you did, in your own words, somewhere the organisation reads, the update, the readout, the document, the review packet. Whoever writes the first accurate description of a piece of work owns its metadata. Be that person. This is not bragging. It is refusing to outsource your authorship.

Translate the work into their protocol, not yours. Your work is meaningful to you in the language of the work, the elegance, the difficulty, the thing you solved that nobody noticed was hard. It is meaningful to leadership in the language of what they are measured on, risk removed, revenue protected, time recovered, a fire that did not happen. These are the same facts in two encodings. If you only ever transmit in the first, the system has to transcode you, and it will do it badly, lossily, and not in your favour. Do the transcoding yourself.

Be in the room, or send a proxy who will represent you accurately. Work gets weighted in rooms. Some of those rooms you can be in. Be in them. Some you cannot, and for those the question is not “will the work speak for itself”, it won’t, but “who in that room will speak for it, and have I handed them the accurate sentence to use.” Absence from the room is not neutrality. It is a decision to be represented by whoever turns up.

None of these requires you to become political in the way you are afraid of. None of them is a lie. All of them are the unglamorous practice of making sure the true description of your work reaches the system before a false one does.

The close

I would like to tell you that once I learned this I started winning every calibration. That I became a fluent operator who never again watched their work attributed to someone fourth-cleverer and first-louder. That is not true either, and you would not believe me if I wrote it.

What changed is smaller and more durable. I stopped believing in the fair evaluator. I stopped treating the description of my work as someone else’s job. And I accepted that the work does not speak, has never spoken, and that “let the work speak for itself” is advice that has only ever benefited the people already holding the microphone.

The work still has to be good. None of this is a substitute for the work, and the people who do distribution with nothing underneath it get found out, eventually, usually. But good is the entry fee, not the win condition. The win condition is distribution, and distribution is just politics with a less frightening name.

You were never going to opt out of it. You were only ever going to do it deliberately, or have it done to you.