The 20-minute version
It's twenty to midnight on a Tuesday and you're at the kitchen table with your laptop.
On the screen is a two-page summary for Thursday's team meeting. Your manager asked for it a week ago. "Just a couple of pages on where the supplier review has got to, so we're all looking at the same thing." You wrote it on Sunday afternoon in about two hours, and at the end of those two hours it said what needed saying.
Since then you've spent four evenings on it. You've rewritten the opening paragraph six times. You changed "significant" to "substantial" in the second section, and tonight you changed it back. You rebuilt the table so the columns line up, decided it looked cramped, turned it into a list, and turned it back into a table. An hour ago you found a sentence on page two that could be read two ways by someone determined to misread it, and you've been working on that sentence ever since.
Your colleague Priya sent hers round on Monday morning. A page and a half, with "Supplier reveiw" in the heading and one figure marked provisional. By Monday afternoon three people had replied, and your manager had written, "Thanks, really useful, let's spend ten minutes on this on Thursday." Nobody mentioned the heading.
You'll send yours at five to nine on Thursday, five minutes before the meeting. Nobody will have time to read it. It will be the better document by a margin only you can see, and Priya's will be the one the meeting discusses.
Count what happened. Ten hours where two had been enough, a short night on Wednesday, and the one thing the summary was for, giving people something to read beforehand, didn't happen.
I'll call the point where a piece of work does its job the line. Polishing is work done past it, where the person receiving the work couldn't tell the difference. The wince is the small jolt when you spot a flaw, or imagine someone else spotting one.
Polishing has tells, by my own observation. The changes go in circles, a word put in, taken out and put back, and a circle means you've passed the line, because real improvements don't reverse. You can't say what you're looking for. You're waiting for a feeling, and it doesn't come.
The polishing makes sense from the inside, because it feels like care. Your standards aren't what's slowing you down, though. Researchers separate two things that travel together, wanting to do excellent work and dreading mistakes and what people will think. The first does little harm and some good. The second does most of the damage, and it's what's doing the polishing. You can keep the first and loosen the second. You decide what done looks like before you start, and when the work gets there you check it once and send it.
Nothing here asks you to care less or turn in rubbish.
Three cautions.
- Perfectionism turns up alongside depression, anxiety disorders, obsessive compulsive disorder and eating disorders more often than chance would predict. If your standards are mostly about food, weight or the shape of your body, this isn't the right help, and the NHS says to see a GP as soon as you can if you think you may have an eating disorder. If your mood has been low for weeks, the NHS advice is the same. See a GP and don't delay.
- If you've had thoughts of ending your life, please get help today. In England that means NHS 111, an urgent GP appointment, or 999 if you're in immediate danger.
- Some work has to be right. If you check drug doses, sign off accounts or service brakes, I won't tell you to check less. Those jobs get accuracy by procedure, and you'll see how further on.
Try one thing before reading on. Think of the last three pieces of work you sent, a report, an email, a birthday card. For each, when was it first good enough to do its job, and when did you send it? If the gaps are small, you may not have this habit. If they're measured in days, read on.
What Sunday did
If someone told you to stop polishing, you'd probably hear it as being told to stop caring. The polishing does give you things. It makes you feel conscientious. Every flaw removed is one fewer thing anyone could criticise, so it feels like closing windows before a storm. It keeps the work on your screen, unjudged, for one more evening. It protects an identity you may have had since school, the careful one, the colleague whose slides never have errors. And once in a while the seventh pass does catch a real mistake, and you remember those catches vividly.
You didn't invent this, either. Thomas Curran and Andrew Hill gathered 164 samples of American, Canadian and British students, 41,641 in all, who had filled in the same perfectionism questionnaire between 1989 and 2016. Average scores rose steadily on all three parts. Students increasingly felt others demanded perfection of them, demanded more of others, and demanded more of themselves, and the biggest rise was the sense that other people expect you to be perfect. It covers students only. But if perfectionism were simply temperament, it wouldn't rise across a generation. Something in how people are raised, schooled and graded is teaching it, whether praise for results and never effort, a parent who spotted the one B among the As, or exams that dock marks for every slip. Those lessons made sense where you learned them. You've gone on applying them in places where nobody is marking.
Now the belief that holds the habit in place. I do good work because I'm a perfectionist. The long evenings are the price of the quality.
That treats perfectionism as one thing. For about thirty years, researchers have found it behaves like two.
Joachim Stoeber and Kathleen Otto reviewed the research in 2006 and described two dimensions that turn up across many questionnaires. Perfectionistic strivings are very high personal standards and working hard to reach them. Perfectionistic concerns are worry about mistakes, doubt about whether what you've done is good enough, fear of criticism, and a constant sense of a gap between what you expected and what you produced. Pulled apart statistically, they point in different directions. Strivings went with conscientiousness, persistence and good exam results. Concerns went with distress. Their conclusion was that striving for perfection is positive in itself, as long as the person isn't overly worried about mistakes and about being judged.
Andrew Hill and Thomas Curran pooled 43 studies of perfectionism and burnout, with 9,838 people from work, sport and education. Strivings had small negative links with burnout, or none. Concerns had medium to large positive ones. With the overlap removed, people with high standards and little dread were the least burnt out of anyone.
These are correlations, mostly from questionnaires, and not every finding flatters strivings. Karina Limburg and colleagues pooled 284 studies on perfectionism and mental health, and both dimensions were associated with psychological problems, though concerns had the larger effect for most outcomes. Eating disorders were the exception, with both contributing about equally, which is one reason for the caution above. The fair summary is that concerns carry most of the cost, and high standards do best when the dread is low.
Go back to Tuesday night and ask which one was working. Your standards operated on Sunday afternoon. They're why the summary was clear, accurate and well organised after two hours, and then they stopped, because it met them.
The four evenings came from somewhere else. Nobody changes "substantial" back to "significant" at twenty to midnight out of love of excellence. You did it because of the sentence someone might misread, the manager who might frown, the colleague who might think you careless. That's the dread. It wears the standards' uniform, which is why it's been so hard to argue with.
So this is what you'd be giving up, on my reading of the research. Keep the standards. Nobody here is asking you to lower them. Turn down the dread, and the Sunday two hours are untouched. What goes is the ten hours after.
Surely, though, the extra passes improve the work a bit?
Joachim Stoeber and Michael Eysenck gave 96 students a proofreading task with as long as they liked. Students with high perfectionistic standards weren't found to catch more of the real errors. They flagged more things that weren't errors at all, and their accuracy per minute was lower. People with perfectionistic standards, the researchers wrote, tend to find fault even when everything is all right. One study of one task, a correlation, and I wouldn't hang everything on it. But it matches the circles. A searching eye with nothing left to find settles on something anyway, and "substantial" was never wrong.
Most of the quality of a piece of work arrives early, and this is my own reasoning. The first draft takes it from nothing to rough, the first proper revision from rough to good. After that each pass adds less, until you're trading an hour for a gain nobody else could detect. For six people reading before a meeting, the Sunday and Wednesday versions were the same document.
Dana Harari and colleagues pooled the research on perfectionism at work and found its overall relationship with job performance was, in their word, equivocal. If the extra hours reliably made the work better, that's where you'd expect it to show. And the hours came out of something, the supplier calls you didn't make, the thinking you didn't do about next quarter, the sleep that pays for Thursday.
None of this makes revision a waste. The first revision is often where the real quality comes from. Past the line, though, the hours stop buying anything anyone receives.
Supplier reveiw
Behind most polishing is a forecast. If this goes out with a flaw, someone will notice and think less of me.
So check it against Priya's heading. You noticed the typing mistake. What did you conclude? I'd bet it wasn't that she's careless and can't be trusted. If you thought anything, it was that she'd typed it quickly, and within a minute you were reading her figures. There's no reason to think people read your work more harshly than you read theirs.
Most colleagues skim what you send between two other tasks. They don't see the difference between your fourth draft and your ninth, and often the eye corrects typing mistakes on the way past. What they do notice is whether it arrived when you said, whether they can understand it, whether it helps them with what comes next, and who holds things up. On every one of those, Priya's page and a half beat your two pages, and her mistake cost her nothing. Lateness is visible to everybody. Most flaws are visible to almost nobody.
There's an experiment often quoted here, from 1966, by Elliot Aronson, Ben Willerman and Joanne Floyd. In the usual account, listeners liked a highly able quiz contestant more after he clumsily spilt coffee over himself. I haven't read the paper, and I wouldn't build on one study from sixty years ago. It does point the same way as ordinary experience. People whose work is good and who are visibly human are easy to like.
Some readers are picky, and there's more on them later. Even then, a determined critic finds something in any document, because there's always something. The forecast assumes a state where nothing could be said, and that one more pass will reach it.
The quieter version of the habit costs more. Some things never go out at all.
The job application you worked on for a fortnight and didn't submit. The email to an old friend after their father died, drafted four times, each draft clumsy next to their grief, until so much time had passed that sending anything felt worse than nothing. The holiday photographs still on a memory card from three summers ago. The belief underneath is that a flawed thing is worse than no thing. Holding it back keeps the excellent version alive in principle, and guarantees a result of nothing.
Your friend didn't need a well-written email. They needed to hear from you, and three awkward lines in the first week would have done it. The employer can't hire someone who didn't apply. You were comparing your draft with a perfect draft. The real comparison was with silence, and the draft would have won easily.
Simon Sherry and colleagues found that psychology professors who demanded more perfection of themselves published less and were cited less, even allowing for conscientiousness. One study, and a correlation, but the extra care wasn't producing masterpieces to justify it.
If there's a message to a person that you've been failing to write well enough, send a short plain version today. "I've been meaning to write for ages and kept not finding the right words. I'm so sorry about your dad. I'm thinking of you." It will land better than you fear, and far better than the fifth unsent draft.
You might hope that at least finishing brings satisfaction. Mostly it brings relief, and within a day you've thought of what you should have changed. Roz Shafran, Zafra Cooper and Christopher Fairburn, in a clinical analysis drawn from people with severe perfectionism, suggest why. When a person's sense of worth depends too heavily on meeting demanding standards they've set themselves, falling short brings self-criticism, and meeting the standard brings no celebration. They decide it can't have been demanding enough, and raise it. It's a theory, and you may be well short of that severity, but the mechanism is recognisable at ordinary strength. Fail and you've failed. Succeed and the test was too easy.
That tells you something about the wince. It presents itself as information about the work. But if you winced at the Sunday version and the Wednesday version, and would wince at a version written by the best writer in the company, then it's firing at anything, and it can't be read as a fault-finder. A spelling checker that underlined every word would tell you nothing about your spelling. So when it comes mid-pass, ask one plain question. Is there a specific fault here that a reader would notice and that would matter to them? If so, fix it. If you can't name one, feel the wince and send the work anyway.
The habit also marks in two categories, right and wrong, with nothing for good or fine or did the job. Its oddest effect is that it makes you do less. If a job can only be done perfectly, then whenever perfect isn't available, nothing is what's left, so the spare room stays untidied and friends come round twice a year.
Mark in grades instead, one to ten, asking what this particular job needs.
- A note to the window cleaner needs a three. Legible, with the right day.
- An email arranging a meeting needs a five.
- A summary for your team needs a seven. Clear, accurate, readable, nothing that would mislead.
- A submission to a regulator or a wedding speech might need a nine.
Hardly anything needs a ten. "Make it perfect" gives you no stopping point, and "make it a seven" does. A seven delivered to a job that needed a seven is a complete success, and the three points you didn't add were never wanted. Someone who gives everything a ten's worth of effort isn't showing high standards. They're showing they can't tell jobs apart.
Grades apply to other people's work too, which the habit soon reaches. You rewrite the colleague's letter, reload the dishwasher, fix the junior's draft overnight without explaining. Each time it feels efficient and even kind. From their side, they learn their efforts get redone, and they step back, and you end up carrying more and wondering why nobody pulls their weight. The dishwasher needs a four. If the plates come out clean, the job's done, whichever way the bowls face. If a colleague's letter is a seven in a different style from yours, it goes out as it is. If it really is a five, tell them the two things that would make it a seven and let them make the changes.
Add it up and the bill is long. Five hours a week of polishing is more than two hundred and fifty hours a year, as rough arithmetic. Then lateness, unsent things, unstarted things, exhaustion. And there's enjoyment. Most people with this habit began by loving some kind of work, and the polishing turned each piece of it into a test.
The line
"Without it I'd get sloppy. The dread is unpleasant, but it's my quality control." This is the fear that keeps people polishing long after they've seen the cost.
Your standards keep your work good, and they don't leave when the dread is turned down. A wrong figure will still bother you. If fear powered the quality, people with high standards and low fear would do worse than the frightened ones. They don't.
Fear is also a poor inspector. The proofreading students caught no more real errors. An anxious checker rereads the same paragraph four times without taking it in, fixes things that weren't broken, and is too tired to see straight because they were up till midnight the night before.
Look at how accuracy is achieved where it truly matters, as general knowledge goes. Airlines, operating theatres and accounts departments rely on procedure. A written checklist, a second pharmacist checking the dose, a colleague who didn't write the figures reviewing them. The check is specific, it has an end, it's done once, often by fresh eyes. When the checklist is complete, the plane takes off.
So here's the method, in two halves.
Before you start anything that would normally tempt you to polish, take two minutes and write four lines at the top of the document.
- Who it's for and what they'll do with it. "The team, six people, to read before Thursday so we can decide whether to drop the second supplier."
- The grade. "A seven." If you want to write nine for everything, ask what would happen if it were a seven. Usually nothing.
- What done looks like, in three or four items someone else could check. "States where each supplier stands. Gives the two prices we're waiting for. Says what I recommend and why. Two pages at most." None of them is "reads beautifully" or "couldn't be criticised", which can't be checked, and that's why they kept you up.
- A time budget and a sending time. "Two hours on Sunday. One check on Monday morning. Sent by ten." Make it earlier than the real deadline and tell the recipient. A promise to a person is harder to let slide.
For small jobs the grade alone will do. "This is a four." Write it once, read it once, send it. For creative work with nobody waiting, invent a reader and promise them a date, and decide in advance how many drafts a piece gets before it goes out. And where you can, show the recipient a rough outline early. It feels exposing, and it's the cheapest guard against ten hours of polish on the wrong question.
Then, at the end, one check from a list, and send.
Five or six questions cover nearly everything that could really embarrass you. Are the names spelt right? Are the numbers and dates correct against the source? Does it answer what was asked? Is the attachment attached? Is it going to the right people? Is there anything I'd regret if it were forwarded? A pass is an open search for anything that could be better. A check asks questions that have answers.
Leave a gap first if you can, an hour or overnight, so you see what's on the page and not what you meant. Fix faults and leave preferences alone. The question is never "could this be better", because the answer is always yes. It's "is this wrong, or unclear to the reader". Go through the list once. Then send it in the same sitting, and don't reread it. Rereading sent work has no function except giving the wince another go.
The first few times you'll press send with the wince still going, and for an hour or so you may feel as if you've left the house without checking the door. That's the fear protesting at not being obeyed. It says nothing about the work, and it's weaker each time.
When a real mistake gets out, and some will, as they do now, the response is dull. "I've spotted an error in the table on page two, corrected version attached, apologies." Then ask whether your list would have caught it, and if not, add a line. Five passes wouldn't have caught it either. You remember the rare catches, and you never counted all the mistakes that got past the polishing anyway.
Treat the first month as an experiment. For each piece you send at the line, note what you feared would happen, and a week later what did. After ten pieces, read it through. I expect the feared reactions to be almost entirely absent. If people really did notice, raise the grade for that kind of work, and you'll have learned something polishing could never tell you. Change is possible even for people far more perfectionistic than you're likely to be. Samantha Lloyd and colleagues pooled eight studies of psychological treatment for clinical perfectionism and found scores fell substantially from before to after, along with anxiety and depression, though eight before-and-after studies is a modest base.
So the decision. I decide what done looks like before I start. When the work gets there, I check it once and send it.
Peter Gollwitzer and Paschal Sheeran pooled 94 tests of "when this happens, I'll do that" plans and found they made people a good deal more likely to follow through. Three to adapt.
When I open a blank document, I'll write the four lines before anything else.
When I catch myself changing something back to what it was, I'll stop, run the check and send.
When I feel the wince, I'll ask what specific fault a reader would notice, and if I can't name one I'll carry on.
A comma at midnight
I can't promise praise or promotion. I'd expect the first fortnight to feel worse at the moment of sending and better in the evenings, and after a month or so sending to have lost most of its sting.
Some situations need more.
If your boss really does pounce on small things, then for that reader the line is higher. Set their work at an eight, put their usual catches on your list, and ask early what they care about most. Don't apply their standard to everyone else. If the pouncing is constant, personal and meant to make you feel small, that isn't perfectionism, and it belongs with someone senior or HR.
If the stakes are high, a grant, a legal document, a safety report, give it a nine. A generous time budget, a longer list, and above all a second reader, who catches far more than your fifth pass because they don't know what you meant to say. High stakes call for a careful process. They don't call for a long stretch of suffering.
If you're a student, read the marking criteria before you start and use them as your description of done. They ask for argument, evidence and structure, hardly ever a flawless sentence, and an essay handed in on time beats a brilliant one capped for lateness. If you keep needing extensions on work that was nearly done, tell your tutor or the university's support service.
When criticism comes, sort it. "The second section confused me" is information, useful and free. "I'd have started differently" is taste, and you can thank them and leave it. If one remark wipes out your week, that's your worth riding on the work, and worth talking over with someone.
If the checking takes hours a day and you can't stop even when you badly want to, or it comes with intrusive thoughts and rituals, see a GP. The NHS describes obsessive compulsive disorder as unwanted, repeated thoughts with repetitive behaviours the person feels driven to carry out, and says effective treatment is available. The same goes if your standards are mainly about eating, weight or your body, or your mood has been low for weeks, or you feel worthless whatever you achieve. In England you can refer yourself to NHS talking therapies. If you've had thoughts of ending your life, get help the same day through NHS 111, an urgent GP appointment, or 999.
And there'll be a week when it all comes back. You'll be tired, or freshly criticised, and you'll find yourself at the kitchen table at midnight putting a comma back. It doesn't undo anything. Do the smallest version. Ask what done looks like for this piece, check it once, send it tonight, and go to bed.
Here's how a Monday might go some months on. An example, not a promise.
Your manager asks for a summary on the contract renewal for Thursday. You open a blank document and type four lines. "For the team, to decide between one year and three. A seven. Three options with costs, risks of each, my recommendation, under two pages. Draft today, check tomorrow at nine, send by ten." You message your manager to say it'll be round by ten tomorrow.
You write it in two and a quarter hours. There's a sentence in the risks section you aren't fond of. You feel the wince, ask what a reader would trip over, can't name anything, and close the laptop at half past five.
At nine on Tuesday you open it with your list beside you. One of the costs is wrong, the two-year figure where the three-year should be. You catch it because "numbers against source" is on the list and you're fresh. You fix it and send at twenty to ten, and don't read it again.
By lunchtime one colleague has a question about the break clause, and another cheerfully mentions you wrote "principle" for "principal". "So I did, thanks." A second's wince, and nothing else. On Thursday the meeting spends fifteen minutes on your summary, because everyone has read it.
The wrong figure got caught by a list and a fresh eye at nine in the morning. The old you might have caught it on the fourth evening, tired and looking at commas, or might not.
So take the next piece of work that comes to you and write the four lines before you start. Tell someone when they'll have it.