Writing It Up

Most test results live in a dashboard nobody revisits. A short written record is what turns a series of individual tests into accumulated knowledge, and it stops someone re-running the same test in eighteen months.

COPY THIS PROMPT - The Test Readout

Write up my A/B test result as a one-page record.

[paste your hypothesis, decision table, result and guardrails]

Use this structure:

WHAT WE TESTED AND WHY

One paragraph. The change, and the evidence that prompted it.

WHAT WE EXPECTED

The hypothesis, including the effect size we predicted.

WHAT HAPPENED

The result with the range. Guardrails. Duration and sample.

State plainly whether we detected an effect.

WHAT WE DECIDED, AND WHY

Reference the pre-declared decision table. If we departed

from it, say so and say why.

WHAT WE LEARNED ABOUT OUR USERS

The most valuable section. What does this tell us that we

didn't know? What belief has it changed?

WHAT THIS DOESN'T TELL US

Honest limitations. What can't be concluded from this.

WHAT WE'D DO DIFFERENTLY

Anything about the test design worth remembering.

Keep it to one page. Write it so someone finding it in a year

understands both the result and the reasoning.

Write Up the Losses Especially

Losing and flat results are the ones most likely to go unrecorded and the ones most valuable to the next person. A written record saying "we tried this in March, here's what happened, here's what we think it means" saves someone three weeks eighteen months from now. Teams that keep this archive compound their learning; teams that don't repeat themselves.