A performance report needs a change log because tracking, consent, site, and campaign settings can change what the numbers mean. Without it, a break in measurement may look like a sudden business trend.
Record changes that affect comparability
Log the date, system, old and new state, reason, owner, expected measurement effect, and verification evidence for changes to:
- event name, trigger, parameters, or deduplication;
- consent banner, tags, or analytics configuration;
- landing URL, form, checkout, or messaging path;
- campaign budget, audience, objective, or optimization event;
- CRM stage definitions, imports, or sales follow-up process.
Do not store secrets or personal customer data in the log. Link to the approved change request or test record.
Example: consent update
If a consent update changes which analytics events can fire, note the release date and the affected regions and event types. Compare pre- and post-change reports with that measurement break visible. Do not “correct” the missing historical events with an estimate unless the method is defensible and clearly labeled.
Use the log when interpreting a trend
Before calling a drop or surge, check whether a logged change occurred in the same period. If definitions changed, create a new series or annotate the discontinuity. Keep a rollback owner and verification step for important tracking releases. Review the log after each monthly reporting cycle and add any undocumented changes while they are still explainable.
Optional AI-assisted work brief — Change log entry generator
Document each tracking or reporting change with timestamp, prior and new behavior, reason, affected properties, owner, validation evidence, and comparison caveat. Link a test record or screenshot without including personal data. Mark periods that should not be compared directly. Do not silently rewrite historical definitions.
Worked example — Annotate a tracking repair
Record that the contact button began firing twice after a template update, the repair date, affected pages, test record, and reporting periods before and after the fix. Do not merge both periods into one trend without annotation. A change log lets an analyst distinguish a measurement discontinuity from a real change in visitor behavior.
AI disclosure: AI tools assisted with drafting. This change-log method is illustrative.