Randomized experiments and holdouts
Compare treatment and control groups when a valid design is feasible.
Marketing measurement glossary
Incrementality in Marketing
Incrementality is the outcome caused by a marketing intervention: what happened with the intervention minus what would have happened without it.
Why incrementality matters
A conversion occurring after an ad does not by itself establish that the ad created it. Incrementality shifts the question from observed credit to the additional outcome that marketing may have caused.
The counterfactual concept
Observed outcome
What happened while marketing ran.
Counterfactual outcome
What would have happened without that intervention.
The counterfactual is not directly observable; it must be estimated with a method and assumptions.
How incrementality is measured
Geo experiments
Use geographic variation and a control strategy to estimate a campaign effect.
Causal methods and MMM
Use observational designs and explicit assumptions to estimate effects when experiments are limited.
Incrementality versus attribution and correlation
Attribution assigns credit according to rules or models of observed journeys. Correlation describes how variables move together. Incrementality is a causal question about the difference made by an intervention. These are not interchangeable claims.
Illustrative example
A team pauses a campaign in a randomized holdout region while continuing it elsewhere. A measured difference, after accounting for the study design, can estimate an incremental effect. That result still has uncertainty and only generalizes as far as the experiment permits.
How MMM relates to incrementality
A Marketing Mix Model can estimate channel relationships from aggregate historical data, but observational MMM does not automatically solve causal identification. Experiments can help calibrate or challenge model assumptions. Hypermacx MMM treats this distinction as part of decision interpretation.
Limitations and common mistakes
Counterfactual estimates depend on design, data and assumptions. Avoid calling all attributed conversions incremental, treating methods as equivalent, or applying a result beyond its evidence. Marginal ROAS can support allocation discussions, but it should be interpreted with the same care.
A decision-oriented measurement practice
Incrementality is one part of a wider evidence system. Marketing Intelligence connects measurement, interpretation, scenarios and learning rather than treating one estimate as a complete answer.