How to Measure Incremental ROAS

Reported ROAS vs incremental ROAS

Reported ROAS = attributed revenue divided by spend, where attributed is decided by the platform's own rules. Incremental ROAS = caused revenue divided by spend, where caused means the orders that would not have occurred without the channel.

A channel can show 5x reported and 1.4x incremental. Same spend, completely different decision.

The incremental ROAS formula

Incremental ROAS = (revenue with the channel - revenue that would have happened without it) / spend.

The hard term is the counterfactual - revenue without it - which you never directly observe. Measuring incrementality is, in practice, the work of estimating that counterfactual credibly.

A worked Shopify example

Say a brand spends €10,000/month on a retargeting channel reporting €51,000 in sales (5.1x). Much of that audience already intended to buy. Estimating the counterfactual from sales history shows ~€14,000 would not have happened otherwise.

€14,000 / €10,000 = 1.4x incremental ROAS. The reported number said scale it. The incremental number said it is barely paying for itself.

Two ways to measure it

1. Holdout/geo tests - suppress the channel for a group and compare. Accurate but slow and costly (see the geo testing guide).
2. Causal modeling on historical data - estimate the counterfactual from variation already in your Shopify + GA4 history, in minutes, with confidence intervals. This is what Causality Engine does.