Sequencing the second move

The situation

The pilot went well. Everyone says so. Six weeks later the second request has been sitting in a finance queue since the last cycle, and nobody is arguing against it — they are simply not funding it. Ask why and you get a shrug: the person who ran the pilot has moved to another team, the number everybody quotes came from a spreadsheet that person built, and no one wrote down what "well" was supposed to mean before it started. The pilot did not fail. It just did not leave anything behind that a finance director can act on.

The principle

A first move is chosen for the evidence it leaves behind rather than the value it captures, because evidence is what pays for the second move.

A first move that saves two hundred hours and leaves no trace buys nothing. A first move that saves forty and leaves a measure your finance team already tracks, one named owner who will repeat the claim in a meeting they did not organise, and a date on which a decision was actually taken — that one buys the next three. Sequencing is choosing the order in which moves compound. Each move should make the following one cheaper by leaving behind exactly one of three things: a measure people accept, a source of data that is now clean enough to use, or a person who has done this once and can do it again without you.

Five fields hold that evidence. Process — one named process, with how many times it ran last month. Measure — one number that existed before the project, and the report it lives on. Owner — one person, not a committee. Decision date — the day you keep it, change it, or stop it, chosen before you start. Stop condition — what you would have to see to shut it down, written while you still do not care about the answer.

Worked example

Shown here in Gemini.

The process is quote follow-up in a regional distributor's inside sales team — the row that scored highest in the first lesson. The notes are messy, which is normal. Assign the brief:

Job: Turn my notes into a one-page first-move brief.

Inputs: The notes below, and nothing else.

Output: Five headed fields and a closing line, nothing more. Process
(named, with runs per month). Measure (one number that existed before
this project, and the report it lives on). Owner (one named person who
can change the process). Decision date (a calendar date). Stop
condition (what we would see that would make us stop). Closing line:
"Second move, and what the first must produce before it starts."

Standard: Where my notes do not contain a field, write MISSING and one
question that would fill it. Never invent a measure, an owner or a
date. Write nothing under a field I did not give you.

Three of the five came back MISSING on the first pass:

Process. Quote follow-up, inside sales. 380 quotes issued last month. Measure. MISSING — which existing report shows quotes converted to orders, and who publishes it? Owner. MISSING — who can change the follow-up sequence without a steering group? Decision date. MISSING — on what date will you keep, change or stop this? Stop condition. Two customers complain about follow-up tone.

That is the brief working. Nobody enjoys reading three MISSINGs, and every one of them would otherwise have surfaced in month four, in front of the people you least want to explain it to.

The closing line is what does the sequencing. Write the second move before the first starts, and name what the first has to produce for the second to begin. If you cannot name it, the first move is a demonstration with a budget code, and you already know how to spot those.

In your tool

A brief that stays in a chat window never gets read. This is about getting it into a document.

Product surfaces checked 2026-08-04.

Try it

Take the process you picked in the first lesson and fill the five fields. Five minutes. Write MISSING where you cannot answer honestly, and do not go looking yet.

You are done when the measure names both a number and the report it already appears on, and when the stop condition is specific enough that somebody who disagrees with you could apply it and reach the same verdict. "It is not working out" is not a stop condition. "Rewrite rate above one in three after four weeks" is.

Common failures

• A measure that was invented for the pilot. If the number did not exist before the project, the project cannot fail — and nothing outside the room accepts it. Pick something already printed on a report somebody else owns, even if it is a blunter number than you would like.
• A committee as owner. Four names on the owner line means the process cannot be changed between now and the decision date, so the decision date will slip, and the evidence will never be produced. One name, or pick a different process.
• Writing the stop condition after the first results arrive. Then it is not a stop condition, it is an explanation. Written up front, it is the single line that makes the whole brief credible to people who have watched pilots run for two years without ever being stopped.