A practical way to review measurement context for an early-stage startup: an incident review
When an early-stage startup has an open tracking incident, change one thing at a time. First write down the symptom, the affected record, and the time it appeared.
Read the symptom
Consent is part of the measurement context. A tag may not run, a signal may be modeled elsewhere, or two reports may contain rows collected under different choices.
Inspect the path
- Record the consent state for the test visit.
- Check which tags fire before and after the choice is made.
- Compare the browser event with the server side record.
- Write down what the system cannot observe under each state.
During the incident
Treat the consent state as part of the evidence instead of hiding it from the report. Keep the test result with the record it describes so the next review starts with evidence, not a guess.
If you need a persistent record of the path, see ClickTrail.