Watch Replays Behind PostHog Funnel Drop-Offs
Use the funnel to locate a meaningful drop-off, then inspect available recordings from that segment. Replays can reveal plausible friction, but a handful of recorded sessions does not represent every non-converter.
Before you start: Have a validated funnel, session replay enabled for an appropriate audience and permission to view recordings. Confirm privacy rules and recording coverage.
Worked example: Users reach an invite form but do not submit. Review sessions for permission errors, unclear labels and interrupted work, then quantify the strongest hypothesis with events.
Step-by-step guide
Validate the funnel first
Confirm step definitions, order, window and exclusions. A tracking error can look like a user-experience problem.
Choose a specific drop-off
Select the transition with meaningful volume and relevance. Record the audience and period before opening recordings.
Open available recordings
Use the supported funnel drill-down or match the segment in replay filters. Note missing recordings rather than assuming every drop-off has video.
Use a review rubric
Record observable behavior, errors and unresolved context. Separate what the session shows from guesses about the person’s intent.
Compare with converters
Inspect some successful sessions under similar conditions. A behavior seen in both groups may not explain the failure.
Validate the hypothesis
Count relevant errors or run a focused test. Save evidence links and limitations before proposing a change.
Common mistakes
Only dramatic sessions
Selecting memorable failures exaggerates a pattern. Use a consistent sampling rule.
Replay proves intent
A pause does not explain motivation. Record the observation and uncertainty.
Missing coverage ignored
Consent and sampling affect available recordings. State the observed subset.
Tips
Write the review rubric before watching sessions.
Inspect successful and unsuccessful journeys.
Keep sensitive details out of shared synthesis notes.
Sources and verification
Based on product documentation checked on 2026-10-24. This is a documentation-based guide, not a hands-on benchmark. Features and availability can change.