AI Product Personalization
Classify users by product behavior, personalize in-app surfaces and emails per segment, then let an AI agent autonomously optimize toward the retention local maximum.
npx gtm-skills add product/retain/ai-personalizationOutcome
>=40% of test group engages with the personalized surface
Leading Indicators
- Personalization surface engagement rate
- Personalized vs control retention at 7 days
- Surface dismissal rate
Instructions
-
Run user-behavior-segmentation drill: query PostHog for 21 days of per-user events, compute behavioral dimensions (primary workflow, session pattern, collaboration), classify into 3-4 segments, store as PostHog person properties.
-
Run personalization-rule-engine drill (smoke scope): pick ONE surface (dashboard/home), create PostHog feature flag with multivariate variants per segment, design variant content per segment, add control variant, split 70/30 personalized/control.
-
Create ONE Intercom in-app message per segment. No email sequences at smoke.
-
Instrument events: personalization_surface_shown, personalization_surface_engaged, personalization_surface_dismissed. Verify in PostHog Live Events.
-
Enable feature flag for 50-200 active users via PostHog cohort. Run for 7 days.
-
Run threshold-engine drill: measure engagement rate >=40% in personalized group, compare vs control (>=5pp lift), guard dismissal rate <40%.
-
If PASS, proceed to Baseline. If FAIL, diagnose per-segment engagement and iterate on weakest variant.
Recommendations
Time
6 hours over 1 week
Play-specific cost
Free (within PostHog and Intercom free tiers for small test group)