Home / Projects / Causal Product Analytics Suite

Causal Product Analytics Suite

One integrated suite across launch impact, growth allocation, and performance diagnosis.

This hub combines three connected analyses used to make high-confidence product decisions: identifying true feature impact, allocating growth spend by incremental value, and diagnosing post-launch metric declines without jumping to false conclusions.

Causal Inference Propensity Matching Difference in Differences Uplift Modeling Media Mix Modeling

Included Studies

Decision-framework diagram. A product or growth decision branches into three studies: feature evaluation (did it really lift retention?) using PSM and DiD, growth optimization (where should spend go?) using uplift and media-mix models, and root-cause analysis (why did the metric drop?) using DiD and cohort-mix. Each yields a causal outcome, and together they feed a decision-first analytics call — ship, no-ship, or reallocate, with evidence.

One question, three causal studies, one evidence-backed decision — the suite as a repeatable ship / no-ship / reallocate framework.

Feature Evaluation ⏱️ 5 min read

When Engagement Lifts Mislead

Used propensity score matching to separate selection bias from feature effect and avoid shipping a change that did not improve retention.

Open Study →
Growth Optimization ⏱️ 6 min read

Optimize for Incremental LTV, Not ROI

Combined media mix and uplift modeling to identify channels that create durable incremental lifetime value rather than short-term attribution wins.

Open Study →
Root Cause Analysis ⏱️ 11 min read

Was the Feed Update the Culprit?

Applied agentic root-cause analysis and DiD to isolate cohort mix and seasonality as the primary drivers behind a post-launch engagement decline.

Open Study →

Methods Used Across the Suite

Counterfactual impact estimation

PSM and DiD to estimate incremental outcomes under controlled comparison logic.

Channel-level growth optimization

Uplift and media mix models to rank channels by true long-term value contribution.

Decision-first analytics design

Metric frameworks and guardrails aligned to ship/no-ship product decisions.

Executive-facing synthesis

Translated model outputs into action-oriented recommendations for stakeholders.

Key Takeaway

Why this suite matters

The three studies work together as a repeatable framework for product and growth teams: evaluate impact causally, allocate investment by incremental value, and diagnose declines with evidence before making expensive rollbacks or scale decisions.