AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

modeling-activation-metrics

Build reusable activation models — an activation-rate metric and a per-user/per-account activated flag — on either PostHog data-warehouse views (HogQL) or an external dbt project. Use when the user wants to define, model, or measure activation, the "aha moment", onboarding success, or which early actions predict a user sticking around. The core idea this skill enforces: activation is NOT a single assumed event — it is a retention-validated combination of early actions, chosen by balancing reach (enough users hit it) against predictive power (those who hit it retain much better). Covers finding candidate actions, validating them against retention lift, count thresholds and action combinations, per-product and B2B group-level activation, and modeling the winning definition as a durable activated-flag + activation-rate model. Read modeling-warehouse-foundations first; composes modeling-product-usage-metrics for the retention validation.

Original source
posthog
License
MIT
Pinned revision
cac7c80a9e5b73484d18eda1bffb47297c938e0c

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

This skill also requires local files or dependencies. Remote reading alone may not be enough.

Ask your assistant

Use Agentic Memory skills to read posthog--modeling-activation-metrics at revision cac7c80a9e5b73484d18eda1bffb47297c938e0c, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 2

references/activation-method.md

Finding and validating an activation definition

Activation is discovered empirically, not declared. The model you build is only as good as the definition behind it, so run this method first.

1. Candidate actions

From read-data-schema, list early actions a new user could take in their first session/week: core feature uses, setup steps, collaboration actions. Include count-based candidates ("did X ≥ 3 times"), not just "did X once".

2. Retention lift

For each candidate, split new users into did it early vs didn't, and compare their retention N weeks later (use modeling-product-usage-metrics). Lift = retention(did) − retention(didn't). A good activation candidate has large, stable lift.

Beware confounds: an action can correlate with retention without being causal (e.g. "visited settings" might just mark already-engaged users). Prefer actions that plausibly deliver the product's value.

3. Reach × predictive power

Score each candidate (or combination) on two axes:

  • Reach — share of new users who hit it within the window. Too low and the metric describes a tiny elite.
  • Predictive power — the retention lift above.

Pick the definition that keeps reach acceptable (rule of thumb: a large minority to a majority of good-fit signups can hit it) while maximizing lift. Combinations ("created a dashboard AND invited a teammate") and thresholds ("≥3 queries") usually beat any single one-time action.

4. Definition shape

The output of this method is a concrete, testable definition:

A user is activated if, within N days of their first event, they [criteria].

Examples: "within 7 days, ran ≥3 queries"; "within 14 days, created a project and invited ≥1 teammate". For B2B, apply the same at the account grain: the account activates when any user meets the criteria.

Record N and the criteria explicitly — they are the model's parameters, and they'll be revisited as the product changes. If a governed activation metric already exists in the semantic layer (see foundations governance.md), reuse it instead of inventing a new one.