Three templates or one? Dev can build either in a day. Maya just wants one more look at the data first.
←→ to move through the storySwipe or use the arrows to move through the story
One More Cut Of The Data — Volume I, Understanding Humans Before Products · Synthesis · Failure & recovery · Practitioner. Maya keeps requesting another analysis on a cheap reversible decision until the cost of the delay quietly exceeds the cost of being wrong
01 · Opening
Three templates or one? Dev can build either in a day. Maya just wants one more look at the data first.
02 · The setup
The question is whether the empty state should suggest three templates or one. Dev can build either in a day.
Dev: Either version is a day. Changing my mind afterwards is another day.
03 · The setup
Maya: Before we pick, can we see it split by first job? Lesson learned the hard way.
04 · The evidence
Week one. The cut arrives and it is genuinely interesting and does not decide anything.
Empty state, by first job — Reporting job — prefers three: 54%; Spreadsheet job — prefers three: 49%; Exploring — prefers one: 52%; Confidence interval: wide
05 · The evidence
Maya: That is nearly a coin flip. Can we look at it by tenure as well?
06 · The evidence
Week three. Four cuts done, two more requested, and the templates are still not built.
Dev: I have been ready for three weeks. The build is one day and we have spent fifteen.
07 · You make the call
The story does not tell you first.
A. The team analyses too much and should move faster generally
B. The cuts were poorly chosen
C. The depth of investigation was never matched to the cost of being wrong
08 · What happened
Dev: If we shipped the wrong one, what happens?
Maya: We change it the next day and nobody notices.
09 · What happened
Maya does the arithmetic she has been avoiding, and it takes about a minute.
Maya thinks: Cost of being wrong, one day. Cost of the delay so far, fifteen. I have been careful into a hole.
10 · What happened
The two costs, finally compared — Build either version: 1 day; Reverse the choice: 1 day; Analysis so far: 15 days; Analysis still requested: 6 days
11 · What happened
Maya: We ship three templates tomorrow and watch it for a fortnight.
Sam: After all that, you are guessing?
12 · What happened
Maya: I am spending a day to learn what four weeks of cuts could not tell me.
13 · What happened
Price being wrong before you price being sure — What does the mistake cost if you make it and reverse it? If that is a day, no analysis costing more than a day can be worth running first.
Check whether the door reopens — Reversible decisions are experiments and want speed. Irreversible ones — a public API, a tier withdrawn from four thousand teams — deserve every cut you can afford, and this framework says analyse harder there, not less.
Count the delay as a cost, out loud — Waiting feels free because nobody invoices for it. Fifteen days of a build sitting finished is a real number, and writing it beside the cost of the mistake usually ends the argument.
14 · What happened
Three templates ship. A fortnight of real behaviour settles it more clearly than any cut had, and reversing it is never needed.
Dev: Two days of work. Four weeks of calendar.
15 · Complete
Depth of investigation should be priced against the cost of being wrong, which sometimes means far more analysis and sometimes means none. Next: the volume's tools, all pointing different ways at once.