# Screenshot experiment worksheet

Blank worksheet from Skir. This is not experiment data or a conversion forecast.

## Before launch

- App / release:
- Platform / experiment ID:
- Hypothesis:
- Control files:
- Treatment files:
- Exact asset change:
- What remains constant:
- Locale / eligible audience:
- Traffic allocation:
- Primary metric and platform definition:
- Minimum practically useful effect:
- Planned review / stopping rule:
- Guardrail metric:
- Reviewer and file approval date:

## Actual observations

- Start / end date and timezone:
- Platform report filename:
- Control sample count / conversion:
- Treatment sample count / conversion:
- Absolute percentage-point difference:
- Relative percentage difference:
- Platform-reported confidence / interval:
- Retention or other guardrail result:
- Known traffic, pricing or release changes:
- Status: NOT RUN / COLLECTING DATA / INCONCLUSIVE / CONCLUDED

## Decision

- Keep control / apply treatment / continue within platform limits:
- Reason supported by the report:
- Decision date and owner:
- Next hypothesis:

Do not infer statistical confidence from a percentage difference alone. Keep inconclusive results explicitly labeled.