Measurement after third-party cookies
Attribution is getting blurrier, not better. How I keep marketing decisions defensible when the data no longer joins up neatly.
Consent mode, tracking prevention and modelled conversions have quietly changed what a dashboard means. Platform numbers now include estimates, GA4 and ad platforms disagree by design, and the honest answer to "where did this lead come from?" is often "several places". That's workable — as long as you stop pretending last click is truth.
Accept three layers, not one number
I run measurement in layers: platform data for in-channel optimisation, CRM data for commercial truth, and self-reported attribution for direction. Each is wrong in a different way, which is exactly why together they're useful.
- Platform: good for bidding decisions, bad for credit allocation
- CRM: the only place revenue actually lives
- "How did you hear about us?" on forms: crude, cheap, surprisingly directional
Make the CRM the scoreboard
Once pipeline and closed revenue are tagged by source, channel and campaign inside the CRM, board reporting stops depending on whichever platform claims credit loudest. It also forces marketing and sales to agree on definitions, which is half the value.
Use incrementality tests for the big calls
Geo holdouts, spend-off tests and clean before/after windows answer the questions attribution can't: is brand search cannibalising organic, does this channel add anything at all, what happens at 30% more budget. They take discipline but they end circular arguments.
Report on a rhythm people trust
Consistency matters more than precision. A stable weekly view of leads, cost per qualified lead and pipeline by channel — plus a monthly commercial readout — earns more confidence than a perfect model nobody can explain.