PostHog investigation

When this applies

Use this after a PostHog funnel, trend, path, or session replay is already in front of you. The connector is https://mcp.usercall.co. The entry tool is create_study. Return to Ask users why for the shared loop.

Signal

Take the PostHog result as given. Name the insight or funnel, the step, the dates, and the counts. A session replay that shows the cursor on a control and then a closed tab is still a signal. It records the path. It does not contain the reason.

Example: Activation funnel, 1–14 September. 1,240 people reached Team invite. 410 sent an invite. Replays pause on the role dropdown, then the tab closes.

Unknown

The open question is why people who reached Team invite left before sending one. The role list, the email field, and a reluctance to invite a coworker are different causes. PostHog has not separated them.

Interview affected users

Interview people who reached Team invite and did not send an invite.key_research_goal is the question. business_context carries the PostHog signal so the interviewer does not re-ask for the chart.

create_study({
  key_research_goal: "Why do people who open Team invite leave before sending an invite? Ask what they expected the step to do, what stopped them, and what they did instead.",
  business_context: "B2B workspace. PostHog funnel Activation, 1–14 September: 1,240 users reached Team invite, 410 sent an invite. Replays pause on the role dropdown, then the tab closes. Interview users who reached Team invite and did not send an invite.",
  target_interviews: 8,
  duration_minutes: 12
})

Send interview_link to that cohort. If the same event is already arriving in UserCall, a Research Trigger can invite them in product later. That trigger is created paused. A person opens activation_url before anyone is interviewed. The investigation still starts at create_study.

Evidence

Poll get_study_status with the returned study_id until status is complete. Then:

get_study_results({
  study_id: "<study_id>",
  format: "summary"
})

Put the themes beside the funnel step. Use format: "full" only when you need a quote. The summary is what you report back. The funnel counts stay the measure of how many people the theme might cover.

Stop here

If PostHog already shows a broken event, a flag that never rolled out, or a step with zero pageviews because the control is missing, report that defect. An interview will not explain a missing button.