Standard Definition
An attribution model is the methodology by which a conversion is assigned to one or more touchpoints from the customer journey. Classical models: Last-Click (100 percent to the last touchpoint), First-Click (100 percent to the first), Linear (evenly across all touchpoints), Time-Decay (more recent touchpoints weighted more strongly), Position-Based (40 percent first, 40 percent last, 20 percent middle). Since 2021, Google has increasingly relied on Data-Driven Attribution (DDA) in its advertising tools — a machine-learning-based model that calculates individual touchpoint contributions based on actual conversion paths. In GA4, DDA has been default since 2021. Important: the chosen model massively influences the evaluation of individual channels — the same advertising channel can perform radically differently under Last-Click and First-Click.
What this means in mandate practice
Attribution is one of the most conflict-laden steering questions in performance marketing practice.
First, no attribution model is „right". Every model is an approximation simplification of the actual customer journey. Last-Click favors performance marketing channels at the end of the funnel (Search, Retargeting), disadvantages upper-funnel channels (Display, YouTube, Social). First-Click favors first-contact channels, disadvantages conversion channels. Data-Driven Attribution is analytically cleaner but black-box — those who switch to DDA change the perceived performance of individual channels, which leads to budget shifts.
Second, attribution conflicts between tools are normal. Google Ads, Meta Ads, GA4, and the respective other advertising tools use different attribution models and each see only their own touchpoint share. The sum of individual tool reports thus often significantly exceeds the actual conversion count. In mandates, this is a recurring client question — the honest answer is: each tool has its own limited view; the overall truth comes only from an overarching analytics setup.
Third, through cookieless tracking, attribution becomes structurally less accurate. With the phase-out of third-party cookies and the limitation of even first-party cookies, multi-touch attributions are increasingly modeled instead of directly measured. Google Ads and GA4 use „modeled conversions" to fill data gaps — which increases accuracy but reduces traceability. Practice recommendation: don't hope for an attribution truth, but work with model bandwidths and interpret trends instead of point values.