Define success metrics for this feature

Before you build, so you're not inventing a metric afterwards to look good.

Define how we will know whether this feature worked.

1. Propose one **primary metric**: the single number that would make us say this succeeded. Just one.
2. Two or three **supporting metrics** that explain movement in the primary.
3. **Guardrail metrics**: what must not get worse. Include at least one that protects users rather than the business.
4. For each metric: the current baseline if I gave it, how it is measured, and the smallest change that would be meaningful rather than noise.
5. **How this could be gamed.** Describe the version of the feature that moves the primary metric while making the product worse.
6. Tell me which of these we cannot currently measure, and what instrumentation would be needed first.

Feature and context:

Discovery and evidence. Checking the idea before you commit: research, feedback, competitors, risky assumptions, success metrics.