posthog-onboarding-conversion

# Which onboarding steps turn trials into paying customers? **Question:** Which onboarding actions best predict that a trial becomes a paying customer? **For:** Growth · **Difficulty:** Beginner · **Shape:** a funnel + a join **Data sources:** PostHog funnels (onboarding steps) + Stripe (first successful payment) ## What this produces A saved PostHog insight comparing onboarding actions completed by users who reached a first Stripe payment against those who never paid, ranked by how strongly each step predicts conversion — so the user knows which steps to push everyone toward. ## Workflow First read `references/posthog-workflow.md` for the shared setup: confirm the PostHog MCP is connected, ensure the Stripe source exists (set it up via the secure connect-link flow if not), and learn this project's real schema. Then do the question-specific part below. ### 1. Identify the pieces in this project - **Onboarding steps.** Use `event-definitions-list` to find signup → activation events. Confirm the intended step order with the user; onboarding funnels are product-specific. - **First payment.** Identify each user's first successful payment from Stripe — typically the earliest paid `stripe_invoice` or successful `stripe_charge` per customer. Join to PostHog people on `lower(email)` (see the join gotchas in the shared reference). - **Conversion label.** For each user, a boolean: did they ever reach a first Stripe payment? ### 2. Build and validate the query Two good approaches — use whichever the user prefers: - **Funnel insight split by converted vs not** (most native): build the onboarding funnel and break it down by a "paid" cohort. This shows step-by-step drop-off for payers vs non-payers directly. - **Comparison query** (ranks predictive steps): compute, per onboarding action, the completion rate among payers vs non-payers and the lift. Start from this shape and adapt names, then validate with `query-run`: ```sql -- Completion rate of each onboarding action among payers vs non-payers, with lift. -- Adapt: onboarding event names, the payment definition, and the email join key. WITH payers AS ( SELECT DISTINCT lower(email) AS email FROM stripe_invoice WHERE status = 'paid' ), user_actions AS ( SELECT e.person.id AS person_id, lower(e.person.properties.email) AS email, e.event AS action, min(e.timestamp) AS first_did FROM events AS e WHERE e.event IN ('signed_up', 'completed_profile', 'created_project', 'invited_teammate', 'activated') GROUP BY e.person.id, lower(e.person.properties.email), e.event ) SELECT ua.action, countIf(p.email != '') AS payers_who_did, countIf(p.email = '') AS nonpayers_who_did, round(countIf(p.email != '') / nullif(count(), 0), 3) AS share_of_doers_who_paid FROM user_actions AS ua LEFT JOIN payers AS p ON ua.email = p.email GROUP BY ua.action ORDER BY share_of_doers_who_paid DESC ``` Note the caveat honestly when you report: this is correlation, not proof of causation — steps that correlate with paying aren't guaranteed to cause it. ### 3. Save the insight Save as a funnel insight (if you went the funnel route) or a SQL/HogQL table insight named "Onboarding steps that predict conversion", described with the payment definition and window. Return the URL and tell the user which steps most separate payers from non-payers. ## Self-driving development (offer this) With the steps that predict conversion, the user can spot trials stalling before them. Offer to help set up in-product prompts, a lifecycle email trigger, or an experiment that moves stalled users forward — turning more trials into customers.