Attribution
The argument over who gets credit for the sale — and why the honest answer is rarely one channel.
- Question
- which touchpoints get credit
- Models
- first, last, linear, position, data-driven
- Limit
- correlation, not causation
- Stronger test
- incrementality
Forms & parts of speech
What attribution tries to do
A customer rarely converts from one touch. They see an ad, read a review, search, click an email, then buy. Attribution is the practice of deciding how much credit each of those touchpoints deserves for the sale.
Different models split the credit differently. Last-click gives it all to the final touch; first-click to the first; linear spreads it evenly; position-based weights the ends; data-driven models infer weights from patterns in the data.
Why every model is biased
Each model encodes an assumption, and each assumption is wrong in some way. Last-click overcredits the bottom of the funnel and makes brand and awareness work look worthless; first-click does the reverse.
Deeper still, attribution measures correlation, not causation. It credits touchpoints on the path but cannot tell you which ones actually changed the outcome. That question belongs to incrementality testing — holdouts and geo experiments — which attribution alone cannot answer.
A holdout test would have revealed the video's true incremental contribution that last-click attribution structurally hid. Attribution framed the question; only incrementality answered it.
Benchmarks
Attribution is a modelling choice, not a metric with a benchmark. Validate channel value with incrementality tests rather than any single model's credit split.
Ranges are illustrative; every published figure is cited from a named public source or labelled “RGM analysis.”
Synonyms & antonyms
Synonyms
Antonyms
Usage trends
Search interest for this term over the last five years:
Common questions
- What is marketing attribution?
- The practice of assigning credit for a conversion across the touchpoints that preceded it, using models like last-click, first-click, linear, position-based, or data-driven.
- Which attribution model is best?
- None is objectively correct — each encodes a bias. The deeper issue is that attribution shows correlation, not causation, so pair it with incrementality testing.
- Attribution vs incrementality?
- Attribution distributes credit across the observed path; incrementality measures the causal lift a channel actually drove, using holdouts or experiments.
Related tools & calculators
Resources & people to follow
- referenceRGM analysis — attribution model biases
- referenceThink with Google — measurement resources
Curated, non-competitor resources verified per term.