Connecting PostHog to TractionScout

Connect a scoped PostHog personal API key to import product-event evidence for the intended project. Check event meaning, identity, dates, and coverage before using those records to investigate signup or activation. A historical read does not grant permission to change experiments or feature flags.

What the connection supplies

The read connector queries historical product events and records available event counts, subject and session references, page and acquisition context, and observation times. Its provider catalog advertises product events, historical event counts, and activation signals. It does not automatically import a saved PostHog funnel definition or create events that were never captured.

PostHog's API documentation distinguishes project keys used for capture from personal keys used for authenticated reads. In TractionScout, use a personal key with query:read for historical import. project:read is needed only when discovering projects. Review the live connection form for your deployment's exact requirements.

Connect the right project

1. Open the intended PostHog project and create a scoped personal API key. Limit it to the read capability and project you intend to analyze.
2. In TractionScout open Configure → Data Sources, choose PostHog, and paste the key in Personal API key. Store credentials through this form, not in a chat, public page, or source file.
3. Set API host to the private API host for your account's region or installation. PostHog Cloud uses https://us.posthog.com or https://eu.posthog.com; a capture/ingestion host is not the query host.
4. Supply the numeric Project ID when needed. The connector also supports @current when the scoped key can query the current project.
5. Review the available history and sync settings, save, and run the first sync. Inspect connection readiness and the last successful observation.

Verify before relying on a rate

Choose a recorded completion event in PostHog and verify that the imported evidence covers the expected project and dates. Check it against the source using the same event filters and population. Account creations, distinct event subjects, sessions, and repeated event counts are not interchangeable.

The historical query has a bounded result size. If a sync reaches its row limit, its coverage is partial and represents the newest returned portion; it cannot establish complete history for the requested period. Inspect coverage before comparing old cohorts or treating a missing event as zero.

For a signup-to-activation question, inspect whether stable identities and usable ordering connect the required steps. For paid conversion, add Stripe and verify the observed relationship to the same users or trials. Similar daily totals alone do not establish a cross-provider funnel.

Common setup problems

Queries fail after saving the key. Check the region's private API host, the personal key's query:read scope, and its project access. A capture-only project key cannot authenticate a historical query.

Events appear but activation is unavailable. Verify the completion event and the project's measurement definition. A connection can be healthy while the outcome is unmeasured.

Experiment execution asks for different access. Keep read analysis separate from mutation. Optional experiment capabilities require additional scopes and the existing action authorization; granting an analytics read does not enable a rollout.

Use the evidence

Start with finding where trial signups get stuck or comparing acquisition channel activation. Use measurement methodology to inspect the meaning of a count or rate. To stop syncing, remove the connection in Data Sources and revoke its personal key in PostHog; previously imported evidence follows the project's data retention and deletion controls.