What problem does it solve?
Support tickets about PostHog Surveys — surveys not showing, fewer responses than expected, incomplete-looking responses, or accidental submissions — are usually eligibility or configuration issues rather than SDK bugs, and misdiagnosing them leads to costly wrong advice like rebuilding surveys.
Core Features & Use Cases
- Eligibility pipeline tracing: Walk the web SDK's ordered eligibility checks (running state, linked/targeting flags, wait period, seen flag) to find exactly which gate failed and why.
- Cross-SDK parity table: Verify whether a feature (URL targeting, popup delay, partial responses, auto-submit) is even supported on the customer's platform before investigating it.
- Response data interpretation: Decode partial responses, branching-skipped questions, and optional questions so complete-but-sparse responses are not misread as broken submissions.
- Diagnostic HogQL queries: Run read-only query templates (shown vs sent ratios, flag evaluation, latency analysis) via the PostHog MCP against the customer's project.
- Use Case: A customer reports "only the first question was answered." The skill guides you to check per-question
branching and optional fields in the survey JSON, revealing the blank cells are questions respondents were never shown — an expected complete response, not a bug.
Quick Start
Paste a PostHog Surveys support ticket and ask the AI to diagnose why the survey is not showing or why responses look incomplete.