Analytics
Last-touch attribution is lying to you
The comfortable default
Last-touch is easy to explain in a weekly standup: the last paid click before purchase gets the credit. That simplicity is why it survives long after most media teams know it is wrong.
The problem is structural. Assist channels that introduce demand look inefficient, while retargeting and branded search absorb credit for journeys they did not create. Budgets follow the report, not the truth.
What multi-touch actually changes
When you spread credit across the journey — linear, time decay, U-shaped, last non-direct, or source-specific lenses — you stop treating every conversion as a single-click story.
The useful move is not picking one “correct” model forever. It is comparing models on the same journeys so you can see where they agree and where last-touch is quietly over-weighting the bottom of funnel.
- Inspect per-touch credit percentages on real journeys
- Compare models and lookback windows without re-ingesting data
- Tie Stripe and CRM closed-won to the same ad paths as click events
Signal first, then credit
Multi-touch on incomplete data is still fiction. If ad blockers and consent gaps erase early touches, every model will still over-credit whatever survived.
Intently attributes on complete first-party signal captured server-side, with nine inspectable models and lookbacks from one day to lifetime, so the debate is about credit math — not missing events.