The cost of guessing
Going back to the brief
A return to the brief is not rework. It is the discovery that the work answered a question nobody asked, and it discards everything built after it.
Decisions this page is about, and where the line actually was
Changing the wording on a customer-facing error message
4 → 1It had been yours all along
Refunding a customer under fifty pounds
3 → 1A written policy you had never been shown
Telling a customer a date
3 → 6Dates are committed by one person here
Reordering your own week
4 → 1You asked twice before believing the answer
- 1Decide it alone and say nothing
- 2Decide it alone and mention it afterwards
- 3Check with one colleague first
- 4Check with the person who assigned it
- 5Needs a decision one level up
- 6Cannot be decided here at all
four decisions · three of them you had more room than you thought · one gave you less · the median gap was three steps on the scale
Four of the nine decisions counted on this site.
A fair pilot starts with a shared purpose and baseline; managers can see how the software works before comparing the results of a new handover practice.
For additional reading outside this collection, see material published by U.S. Department of Labor.
The two pieces of work that went all the way back to the brief
The quarterly numbers
4 stages backIt went back because nobody had said who the audience was. Rebuilt from scratch for a different reader
The reporting page spec
2 stages backIt went back because the purpose turned out to be a different one. A dead prototype for the same thing turned up in a search
- 1the brief
- 2the plan
- 3the build
- 4review
- 5shipped
two pieces of work came back · the median went back four stages · the furthest went back four · two of them went all the way to the brief
Two of the six returns counted on this site, drawn on their own because a return to the brief is a different event from every other kind.
Two of six returns went all the way to the brief. One of them had already shipped.
A return to the brief is categorically different from every other kind. When work goes back one stage, something in the work was wrong. When it goes back to the brief, the work was fine and was about the wrong thing — which means every decision downstream of the misunderstanding is discarded at once, regardless of how good any of it was.
The quarterly numbers
The work was to produce the quarterly numbers. It was done thoroughly, at analyst depth, over several weeks, and it shipped.
The audience turned out to be the board. Not a different level of detail — a different document: different structure, different framing, different things left out, different definition of what a number even is in that context.
Nothing in the first version was wrong. Every choice in it was correct for the reader it assumed. The gap between "the quarterly review" and "the board reads this" is one sentence, and it was the difference between four weeks of useful work and four weeks of work that was thrown away.
The reporting page
The second is a milder version with the same shape. The work got to the build stage before the purpose turned out to be different from the stated one, and a dead prototype for the same thing surfaced in a search once somebody knew what to search for.
Two of the four missing things in one piece of work: the purpose, which caused the return, and the precedent, which would have caused it sooner and cheaper.
Why these are not caught earlier
Because nothing in the normal process looks for them.
Review looks at the work against the brief. If the brief is wrong, the work passes review and fails later, in front of whoever actually wanted it. Both of the returns above were found by the real audience, not by any internal step.
Status updates make it worse rather than better. A weekly update says what was done, and what was done sounds right, because it is right against the assumed purpose. Nobody reading "finished the detailed breakdown by segment" will think to ask who the reader is.
The one check that catches them
Show the intended output to the intended audience, early and in rough form.
Not a status update and not a plan — a sketch of the actual thing: one page of the document, one screen of the page, three numbers in the shape they will appear. Fifteen minutes of work, shown to the person who asked for it.
Both of the returns on this panel would have been caught in week one by that. The board would have looked at one page of the analyst version and said "this is not what we need", which is a four-minute conversation in January and a four-week loss in March.
Why the deepest return often happens last
Both of the returns on this panel were discovered late — one after shipping — and that is not a coincidence. The deeper the misunderstanding, the later it surfaces.
A detail error is caught by the first person who looks carefully, because a detail is visible against the brief. An approach error is caught at the plan stage, if there is one. An error about what the work is for cannot be caught by anybody working from the same brief, which means it survives every internal check and is found by the audience.
That produces a counter-intuitive pattern: the better the internal process, the later these surface. A team with thorough review will catch every detail and ship the wrong thing, confidently, having checked it four times.
It also explains why these returns feel so much worse than their size. A one-stage return happens in private, among people doing the work. A return to the brief happens in front of whoever asked for it, after somebody has presented it, which adds an entirely unnecessary layer of embarrassment to what is really an information failure.
Both of those arguments point the same way: the only check that catches this is outside the team, early, and in rough form. Fifteen minutes with the real audience in week one is worth more than any number of internal reviews in week eight.
Saying it when it happens
When something does go back to the brief, the useful response is not an apology and not a post-mortem. It is one sentence connecting the cause to the cost while both are visible:
This goes back to the outline, about three days, because the reader turned out to be the board rather than the review. Worth saying who the reader is next time.
That sentence is the entire feedback mechanism available to whoever hands work over. Without it they see a piece of work that took longer than expected; with it they see the ninety seconds they did not spend.