The bar says 60%.
It says so because you ticked six things out of ten, or because you've written thirty thousand of a fifty thousand word target, or because you told it in January that the project had five phases and you're most of the way through the third.
None of those means you are 60% finished, and the gap between the number and the truth isn't small. It's often the difference between a project that's nearly done and one that's about to go badly wrong.
What the number is actually measuring
A percentage needs a denominator, and in knowledge work the denominator is a guess.
Thirty thousand of fifty thousand words is 60% of an estimate you made before you'd written anything, at the point of least information you'll ever have about the book. Six of ten tasks is 60% of a task list that was written before you understood the problem, and which is already missing three things you've since discovered.
So the bar isn't measuring progress toward the finish. It's measuring progress toward January's idea of the finish, and the two diverge steadily and invisibly as you learn what the work actually involves.
Real work goes backwards
The deeper problem is that these bars can only increase, and the work doesn't.
A chapter that was finished becomes unfinished the moment you realise the argument doesn't hold. A feature that was done gets scrapped when it turns out to solve the wrong problem. Two weeks of genuinely excellent work can leave you with less than you started with, because what you did was discover that a section has to go.
That's not failure. It's the normal shape of thinking, and it's often the most valuable work in a project. But a bar that only rises cannot represent it. So you delete a chapter and the bar stays at 60%, or you re-add the words with something better and it climbs, having recorded a fortnight of hard thinking as either nothing or as a slight setback in word count.
Any measurement that can't go down is not measuring the thing it claims to.
There's a related distortion in how the bar gets fed. Because ticking things off is what moves it, the easy items get done first — they're worth the same as the hard ones to the number, and they cost a fraction of the effort. Six weeks in, the bar reads 70% and every remaining task is one of the difficult ones. The percentage has been accurate about task count and wildly misleading about time, which is the only thing anybody wanted it for.
The number changes decisions
If this were merely inaccurate it would be a curiosity. What makes it expensive is that people act on it.
A bar at 90% produces optimism about a deadline, and that optimism gets communicated. Dates get promised. Other work gets scheduled after it. The last 10% of a piece of knowledge work routinely takes as long as the first half, because the last 10% is where all the hard remaining decisions live — and nobody plans for that, because the bar said 90%.
There's a subtler version too. A high bar makes you reluctant to do the work that would lower it. If you're at 85%, noticing that the structure is wrong means going back to 50%, and there's a real pull to not notice. The bar starts protecting itself, and it does so at exactly the point in a project where honest reassessment is most valuable.
Deadlines built on that optimism then propagate outward. Someone else schedules around your date, and the cost of admitting the bar was wrong grows every week, until the sensible move — going back to 50% and fixing the structure — has become organisationally expensive as well as personally unappealing.
What's true about the criticism of this
The counter-argument deserves a proper hearing, because progress bars aren't only theatre.
For work that genuinely decomposes, they're accurate and useful. Painting fourteen radiators, migrating four hundred records, filing thirty documents — the denominator is known, the units are equivalent, and the bar means precisely what it says. Plenty of real work is like this, and treating percentages as universally dishonest is its own kind of wrong.
They also do something psychological that shouldn't be dismissed. A visible bar makes a long project feel finite, and finiteness is motivating in a way that an open-ended task isn't. Some people finish things because of a bar, and an argument that ignores this is arguing against a strawman.
The honest position is narrower. Percentages are fine when the denominator is known and the units are alike. They mislead exactly when the work involves discovering what the work is — which is most of what people care about, and all of what they find hard.
Thirty thousand words in
Someone is writing a technical book. The target is fifty thousand words, and by June they have thirty-one thousand. The bar in their project tool says 62%.
In July they realise chapters four and five are making the same argument, and that the book's structure is wrong: it's organised by feature when it should be organised by task. Fixing it means cutting about nine thousand words and rewriting most of chapter six.
They spend three weeks on this. It's the best work they do all year, and it turns a book that would have been mediocre into one that works.
At the end of July the manuscript is twenty-six thousand words. The bar says 52%. By its own account they have gone backwards by ten percentage points and lost a month.
The log for the same period says: found four and five overlap, restructured by task not feature, cut ch5, folded the good bits into four, rewrote six around the new order. That's a legible account of the most important month in the project, and it's available because the record described what happened rather than scoring it.
What to measure instead
Two things, both of which stay honest when the work turns.
What you did, in one line, each time. It can't go backwards because it isn't a score — it's a record, and cutting nine thousand words is an entry like any other. Read back in October it tells you where the project went, including the parts that involved subtraction.
And how long since you last touched it. That number is a fact rather than an estimate. Nobody has to guess at a denominator, it can't be gamed by doing the easy tasks first, and it's equally true whether the work is going well or badly.
Neither tells you how close you are to finished. Nothing can tell you that, which is the actual point — the certainty a progress bar offers is the thing that isn't available, and inventing it costs more than not having it.
If you need a sense of remaining scope, the honest version is a list of what's left, rewritten when it changes, with no number attached. It's less satisfying and it doesn't fit in a dashboard. It also survives the week you delete a chapter, which no percentage does.
Related: The case against streaks · Logging beats tracking · What we left out of goaldrift, and why
Share: