Attribution vs incrementality: when to use which
They answer different questions. Treating them as substitutes wastes budget.
Attribution tells you which touchpoints get the credit. Incrementality tells you which channels would still drive sales if you turned them off. Both matter. Most marketing teams use one of them well and the other badly — usually attribution well and incrementality not at all.
The decision rule
Use attribution for daily optimisation. Use incrementality (geo holdouts, ghost bidding, conversion lift studies) for quarterly budget allocation. Don't try to optimise daily bids based on a quarterly incrementality study.
The cheapest incrementality test
Pause a single channel in three regions for two weeks. Compare to matched controls. If you can't measure the impact, the channel was overrated. Total cost: opportunity cost of two weeks' spend in three regions, which is usually under 5% of monthly spend.
What incrementality usually reveals
- Brand search is largely cannibalising organic — uplift is real but smaller than reported.
- Retargeting often overstates contribution by 30–50%.
- Top-of-funnel video shows up as incremental but invisible in attribution.
- Affiliate frequently shows zero or negative incrementality on existing customers.
What to do with the gap
Don't reallocate overnight. Use incrementality to set 90-day budget targets and let attribution guide week-to-week within those targets. The teams that switch off cannibalising channels too fast usually see short-term revenue dips that nobody trusts.
