Write the misses on the wall too: how the outcome ledger works

"Try a different creative." "Put more budget behind it." With no metric or window fixed in advance, nobody can say afterwards whether it worked. The ledger's rule is simple: fix it before, record it honestly after.

The most expensive waste in growth isn't misspent budget — it's spend you learn nothing from. Thirty actions in a quarter, and at the retro maybe three have a legible cause and effect. The rest become "felt like it helped."

The acceptance contract

Before any proposal is confirmed, four things get fixed: which metric, which baseline window, how long the measurement window runs, and what conditions void the measurement (the confounder guards). Then it executes, and results backfill automatically when the window closes. The point is to turn "find an explanation afterwards" into "commit to a judgment beforehand."

Recorded honestly

• Hit: new-angle content ER +41%, inside the pre-registered +30%–50% band.
• Partial: paid CPC -14% against a -25% expectation — right direction, short of the mark; the gap's main driver fed back into the creative library's tags.
• Miss: goes into the ledger too. Not deleted, not buried.

Not cherry-picking the flattering numbers is what makes the system credible. And because the ledger holds the misses, the next proposal can cite your own measured results instead of somebody else's case study.