Attribution Model

Types of attribution models

Every conversion has a path — a sequence of interactions (ad clicks, organic searches, email opens, direct visits) that led to the final action. Attribution models define how to assign value across that path:

— First-touch assigns 100% of credit to the first interaction. Useful for measuring awareness channels, but ignores everything that happened after discovery.
— Last-touch assigns 100% to the final interaction before conversion. GA4 defaults to this for most reports. It overstates bottom-funnel channels like branded search and retargeting.
— Linear splits credit equally across all touchpoints. Simple and fair, but assumes every interaction has equal influence — which is rarely true.
— Time-decay gives more credit to touchpoints closer to conversion. Reasonable for short sales cycles, less useful for B2B journeys spanning weeks or months.
— Data-driven uses machine learning to calculate each touchpoint’s actual contribution based on conversion probability. Google removed all other models from GA4 in late 2023, making data-driven the default.

Why attribution needs complete data

Every attribution model — from the simplest last-touch to the most sophisticated data-driven — depends on seeing the full journey. When analytics data loss removes 40–87% of touchpoints, the model works on a fragment of reality.

Consider a customer who first discovers your brand through an organic search (blocked by an ad blocker), later clicks a display ad (tracked), and finally converts through a branded search (tracked). A last-touch model credits branded search. A data-driven model credits display. Neither knows the organic visit existed. The channel that actually introduced the customer gets zero credit — and zero budget in the next planning cycle.

Multi-touch attribution needs more than complete data — it needs a persistent identifier linking the same visitor’s touchpoints across sessions, which is exactly the cookie dependency that causes the data loss above. Cookieless analytics closes a different gap: it captures every touchpoint within a session without that identifier, which is why models built on it — like last-click — run on 100% of sessions instead of a consent-biased subset.

Attribution model comparison

Note: GA4 deprecated all models except data-driven and last-click in November 2023. Data-driven attribution requires sufficient conversion volume and — critically — complete data to produce reliable results.