Verify PostHog Experiment Exposure
Check that exposure means a person reached the experience under test, not merely that a flag was fetched somewhere unrelated. Validate exposure recording before interpreting metric differences.
Before you start: Have a draft or test experiment, instrumentation access and documented eligibility. Know whether analysis uses the default flag event or a custom exposure event.
Worked example: An onboarding experiment changes an invite screen. If the flag is evaluated at login, many assigned people may never reach the screen, diluting an analysis intended to measure that experience.
Step-by-step guide
Define meaningful exposure
Write the moment at which the person can encounter the treatment. Keep assignment, eligibility and actual encounter distinct.
Inspect flag evaluation
Find where the application evaluates the relevant flag and what event is recorded. Verify that the flag key and variant values match the experiment.
Test each variant
Use controlled test users to traverse the relevant flow. Check the exposure event and a downstream metric event under the correct identity.
Review exclusions
Verify internal-account rules and handling of people associated with multiple variants. Follow the experiment’s supported settings rather than silently dropping inconvenient data.
Check allocation
Compare observed exposure distribution with the planned allocation. Investigate a mismatch or a missing platform before interpreting an apparent winner.
Record validation evidence
Save the test journeys, event examples and exposure definition. Re-run checks after a rollout or instrumentation change that affects eligibility.
Common mistakes
Assignment equals encounter
A flag evaluated elsewhere may count users who never saw the change.
Metric events before exposure
Those events may not belong to the intended test effect.
Imbalance ignored
Unexpected allocation can indicate tracking or eligibility problems. Investigate before concluding.
Tips
Keep an exposure test case for each platform.
Record when exposure instrumentation changes.
Verify returning users as well as fresh signups.
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.