User Propensity Score Matching

Estimate whether an operational intervention created incremental impact by comparing treated users with untreated users who looked similar before the intervention. Keep the workflow independent of any analytics platform. Operate read-only by default and never write cohorts, tags, tables, or reports back to a source system without explicit approval.

What This Skill Solves

Game teams often compare campaign participants with non-participants and mistakenly attribute pre-existing user differences to the campaign. This skill reduces that selection bias with propensity score matching (PSM), then exposes the evidence needed to decide whether the comparison is credible.

Explain the distinction on first use:

• Propensity prediction asks who is likely to pay, churn, or participate.
• Propensity score matching asks whether an intervention helped users who received it.

Primary users are game operations, growth, monetization, product, and data teams. Common delivery variants include:

1. Campaign, offer, coupon, or bundle effectiveness.
2. Push, inbox, win-back, or churn-intervention effectiveness.
3. Feature, mode, difficulty, or gameplay-change effectiveness.
4. Acquisition-channel, retargeting, support, or compensation effectiveness.