Causal inference checklist for product teams

Use this checklist before turning a dashboard movement into a roadmap decision. It separates correlation, cohort drift, seasonality, and plausible treatment effects.

Define the causal question

Write the treatment, outcome, population, and time window.
Name the decision that will change if the estimate moves.
Record expected direction before looking at the result.

Check identification

List confounders that affect both exposure and outcome.
Confirm pre-treatment covariates are measured before exposure.
Run negative controls or placebo windows where possible.

Stress-test the result

Compare cohorts, segments, and pre-period trends.
Run sensitivity checks for missing data and outliers.
Document what evidence would reverse the recommendation.

When to use WhyLab

WhyLab is useful when the team has event data but needs a repeatable causal review surface before shipping, pausing, or scaling a product change.

Open the causal lab