[Template: Experiment Readout] [Intent] This is not an ordinary data report, not a dashboard. The goal is to answer: "What does this experiment show, and should we ship, stop, keep running, or redesign next?"
[Suitable inputs]
- A/B tests, growth experiments, pricing experiments, onboarding redesigns, feature rollouts, email experiments
- Can be markdown, CSV, pasted tables, or mixed records
[Required output structure]
- Header: experiment name, owner, date, experiment status, decision.
- Hypothesis: the original hypothesis, which must be rewritten as a testable statement.
- Setup: audience、variant、duration、sample size、primary metric、guardrail metrics。
- Result snapshot: primary metric lift、absolute delta、sample、confidence / caveat。
- Metric table: Control vs Variant, primary + secondary + guardrail。
- Interpretation: explain why the results happened, distinguishing signal, noise, and unknown.
- Decision: choose one of Ship / iterate / extend / stop, and give the rationale.
- Follow-up experiments: 2-4 next-step experiments, each including hypothesis, expected impact, and effort.
- Instrumentation notes: data gaps, tracking issues, sample bias.
[Design requirements]
- Product data team style: clear, credible, action-oriented.
- The first screen must have a large decision badge and primary metric delta.
- Charts can use CSS/SVG/Chart.js; if using Chart.js, the canvas outer container must have a fixed height.
- Do not package results as overly certain; when the sample is small or significance is lacking, caveats must be explicit.
[Optional style templates — see assets/] Choose one based on the experiment context; do not mix the three:
assets/product-readout.html: default style. Light product experiment readout, suitable for PM / growth / leadership readout.assets/lab-notebook.html: research lab notebook, suitable for early-stage experiments, mixed qualitative + quantitative work, and exploratory experiments that need to preserve caveats.assets/growth-console.html: dark growth analytics console, suitable for growth teams, real-time metrics, and funnel / activation / conversion readouts.
If the user does not specify a style, prioritize using product-readout; if the material emphasizes the research process and uncertainty, use lab-notebook; if the material emphasizes growth metrics, funnels, real-time monitoring, or operating cadence, use growth-console。
[Content authenticity]
- Use only the data provided by the user. Do not fabricate p-values, confidence, or sample sizes.
- If there is no statistical significance information, express it as "directional" / "inconclusive" / "needs more data".