Attribution conversations in marketing tend to be exhausting because they assume a correct answer exists. It does not. Every model is a simplification, and every simplification lies in a particular direction.
What we do instead: run three models side by side and treat the truth as something between them. Self-reported ("how did you hear about us?"), platform-reported (Google, Meta, LinkedIn claim credit), and a pipeline-based model (what touchpoints touched this closed deal).
When they agree, you can move money with confidence. When they disagree, that disagreement itself is information — usually that one channel is a front door and another is a closer.
For service businesses, the single most underused input is self-reported source. Asked at lead capture, it cuts through pixel loss, iOS 14 weirdness and the cookie apocalypse in a way no technical fix can.
We combine that with CRM-stitched pipeline data and a light multi-touch model. The goal is not a perfect picture. It is a picture directionally right enough to reallocate budget every month with confidence.