One definition
Turn ambiguous KPIs into contracts with formula, grain, owner, and limits.
Analytics engineering toolkit
Metric Contract Studio
Business teams argue about revenue, conversion, and churn because nobody owns a single definition. This studio turns metrics into contracts — validated, scored, and exportable.
Frontend-only demo · localStorage · no backend required
Explainable maturity
Score every metric from 0–100 and show exactly which points were earned or lost.
Ship documentation
Export Markdown, SQL templates, and quality checklists for wiki or PR review.
New · MetricGuard CI
Mudei um JOIN. Nenhum teste quebrou. Minha receita quase dobrou.
See the deterministic gate block a PR where a valid-looking JOIN fans orders out and inflates net_revenue by +66.7% — with row-level evidence.
Five-minute path
From ambiguous KPI to reviewable contract
- 01Open the conversion conflict case — two dashboards, one label
- 02Load demos and open Taxa de conversão (resolved contract)
- 03Inspect score, grain, filters, and incorrect-usage example
- 04Submit for review, approve, and read version history / diff
- 05Export Markdown + dbt YAML as the handoff artifact
Demo set
5 ready contracts
Net revenue, conversion rate, monthly churn, average order value, and user activation — each with formula, grain, validations, and misuse examples.
Why metrics break
Dashboards disagree when grain, filters, and ownership are tribal knowledge. A metric contract makes the definition reviewable — before the board meeting, not during it.
- • Missing owner and refresh cadence
- • Ratio metrics without numerator/denominator
- • No validation rules or misuse examples
- • SQL copied from Slack with no source of truth
Portfolio signal
Governance as a product
Built to demonstrate analytics engineering judgment, not just UI scaffolding.