← All Labs

Incrementality

Flagship
Measurement
01

Customer question

Would these customers have converted anyway?

02

What this solution does

Attribution tells you which ads got credit for a sale. Incrementality tells you whether the ad actually caused the sale, by comparing against people who saw no ad at all.

Specifically, it answers: Of the conversions a platform attributes to advertising, how many actually happened because of the advertising, versus how many would have happened regardless?

03

How it works

A treatment group is exposed to advertising and a control group is deliberately withheld from it (a holdout); comparing conversion rates between the two groups reveals the incremental effect the platform-attributed number does not.

04

Data requirements

  • A genuine, sufficiently sized holdout/control group
  • Randomised or matched assignment to treatment and control
  • Conversion tracking consistent across both groups
  • Platform-attributed conversion figures for comparison
05

Discovery questions

Can a true holdout group be created, or only a geographic/time-based proxy?

A true randomised holdout gives the cleanest causal estimate; proxy designs (geo-based, time-based) introduce more assumptions and potential bias.

How large does the holdout need to be to detect a meaningful effect?

Underpowered tests can fail to detect real incrementality, leading to false conclusions that a channel doesn't work.

How will stakeholders react to incremental ROAS being lower than platform-reported ROAS?

This is usually the most contentious moment in an incrementality engagement, and needs to be set up as an expectation early.

06

Build an experiment

1
07

Generate synthetic data

08

Data readiness

Generate a dataset in the previous step to run readiness checks.
09

Run model / simulation

Generate a dataset first to run the model.
10

Results

Run the model in the previous step to see results here.
11–12

Business ⟷ Technical interpretation

Of the platformAttributedConversions conversions the platform attributed to this campaign, only incrementalConversions appear to be genuinely incremental — the rest would likely have happened anyway. Attribution is not causation.

Shown as a template — the highlighted tokens are filled in once the model has been run.

13

What would you do next?

  • Recalculate media efficiency using incremental ROAS instead of platform-attributed ROAS
  • Run incrementality tests periodically, as true incrementality can shift with market conditions
14

Solution & vendor landscape

Implementation approaches

Platform-native conversion lift / holdout toolsIndependent incrementality testing vendorInternal experimentation via geo-holdout or ghost ads
See the full Vendor Landscape →