Attribution
Perfect attribution doesn’t exist. Useful attribution does.
Every attribution model is wrong about something. Last-click misses brand. Multi-touch over-credits the loudest channels. MMM is slow. The marketers who win don’t find the right model — they pick a stance, name its assumptions out loud, and make decisions from it anyway.
How we work
We build attribution you can defend in a meeting.
The job isn’t producing one number nobody trusts. It’s producing a model whose assumptions you can name, whose blind spots you can map, and whose output the marketing team and the finance team both agree to use the same way.
Where the stakes justify it, we layer in incrementality tests so the model isn’t only theoretical. Where they don’t, we name what we’re guessing at — and revisit it when the cost of being wrong rises.
What we hand off
Five views that survive scrutiny.
- A primary attribution model with assumptions written down, not hidden.
- Last-click, first-click, multi-touch in one view — so you can see where they disagree.
- A decision framework for which view answers which question.
- Incrementality tests where the budget at stake justifies them.
- Reporting that holds up under finance scrutiny — and survives the auditor’s questions.
Three attribution models telling you three different stories?
Thirty minutes, no pitch. We’ll figure out which story to commit to — and what to instrument before you bet the budget on it.