# 0032. No single success metric

- **Status:** Accepted
- **Date:** 2026-08-08

## Context

A project that names one measure — and repeats it in the plan template and the phase review — plans
differently: a criterion that is asked every time stops being optional. The question was therefore put here,
with the candidates this project's shape suggests: booking conversion, occupancy, office throughput, the
accuracy of the rental and billing data.

## Decision

**There is no single success metric. This is a feature-delivery project.**

Work arrives as a specified feature or fix, and "done" means it is delivered, correct and approved — not that
it moved a number. Nothing is added to the plan template or the phase review to ask about one.

## Alternatives

- **Naming one anyway** — booking conversion is the closest candidate. Rejected because a measure nobody
  actually steers by becomes a line everyone writes "n/a" against, which is worse than its absence: it makes
  the template look thorough while carrying nothing.
- **Adopting another project's measure** — organic search visibility was the one on the table. It measures a
  different product. Copying it across would have violated Rule 0 in the most expensive way: a rule that is
  wrong is worse than a rule that is missing, because it is followed.

## Consequences

- Task planning stays scope-driven: the acceptance criteria in the plan are the whole bar.
- Prioritisation between tasks stays the client's call, made outside the repository. There is no written
  criterion to appeal to when two features compete.
- If that changes — if the client starts steering by occupancy or by conversion — this ADR gets superseded
  and the standing question gets added then, when there is something real behind it.