Essential session storage is required to sign in. Optional analytics remain off unless you allow them.
Produce a stakeholder-ready deliverable: either an interactive dashboard (Tableau/Power BI/Looker, or a notebook/Streamlit) OR a written insight report — that answers a real question end-to-end. It must include (a) a defined set of KPIs that map to a goal (no vanity metrics), (b) honest visualizations, (c) a data-storytelling structure (context -> finding -> implication -> recommendation), and (d) a statistical readout of an A/B test or comparison: state the hypothesis, the per-arm rates with HONEST denominators, the result, and interpret the p-value and confidence interval CORRECTLY (no causal overreach beyond what the design supports).
A/B readout uses per-arm denominators; p-value and CI interpreted correctly (procedure-based, not 'prob null is true'); significance vs practical distinguished; no unwarranted causal claims.
KPIs tied to a real goal (not vanity); findings are actionable; leading/lagging distinguished where relevant.
Context->finding->implication->recommendation structure; audience-first; honest, clear charts.
Polished, reproducible deliverable a stakeholder could act on; assumptions and limitations documented.
Use this README structure to make the work reviewable.
DevPath fetches a read-only, size-capped snapshot. Submitted code is never executed.
# Insight report or dashboard with an experiment readout ## Goal Produce a stakeholder-ready deliverable: either an interactive dashboard (Tableau/Power BI/Looker, or a notebook/Streamlit) OR a written insight report — that answers a real question end-to-end. It must include (a) a defined set of KPIs that map to a goal (no vanity metrics), (b) honest visualizations, (c) a data-storytelling structure (context -> finding -> implication -> recommendation), and (d) a statistical readout of an A/B test or comparison: state the hypothesis, the per-arm rates with HONEST denominators, the result, and interpret the p-value and confidence interval CORRECTLY (no causal overreach beyond what the design supports). ## Acceptance criteria - [ ] Statistical correctness: A/B readout uses per-arm denominators; p-value and CI interpreted correctly (procedure-based, not 'prob null is true'); significance vs practical distinguished; no unwarranted causal claims. - [ ] KPIs & insight: KPIs tied to a real goal (not vanity); findings are actionable; leading/lagging distinguished where relevant. - [ ] Storytelling & viz: Context->finding->implication->recommendation structure; audience-first; honest, clear charts. - [ ] Delivery: Polished, reproducible deliverable a stakeholder could act on; assumptions and limitations documented. ## Verification - [ ] Document installation and run commands. - [ ] Record automated checks and their exact commands. - [ ] Add representative output, screenshots, or a short demo where useful. - [ ] Confirm failure paths and known constraints. ## Decision log ### Decision title - Context: - Choice: - Alternatives considered: - Tradeoffs: ## Evidence - Link each rubric criterion to the file, test, or artifact that demonstrates it. ## Known limitations - List what is intentionally out of scope and what should be improved next.