Signal loss after the cookie prompt
Drop-off charts often start at the banner, not at the hero. How to tell a consent cliff from a content problem.
Journal
Notes we would rather publish than bury in a slide appendix.
Modelling consent states in the warehouse
Unknown is a fact. Overwriting it with the last grant of the day is how yesterday’s board pack becomes fiction.
When reject is the majority path
Product still happens for people who refuse analytics. Measurement has to describe that majority without inventing IDs.
GPP notes for UK measurement teams
What a global privacy string changes in a GB stack, and what it does not excuse in your event dictionary.
Designing fallback events
A small catalogue of signals that remain after reject, and the ones that only pretend to remain.