ROAS tells you what a platform may claim, not what your spend causes. Here is how simple holdout tests reveal which revenue would not exist without your ads.
Incrementality is the revenue that would not have existed without your ads. It sounds like ROAS, but it answers a fundamentally different question: attribution measures what a platform may claim, incrementality measures what your spend causes. You measure it by deliberately withholding ads from a group, a holdout, and comparing the difference. That is simpler than it sounds, and it changes how you decide about scaling.
What does incrementality actually mean?
Imagine someone has your product in their cart, hesitates, and buys the next day. Along the way they saw one more retargeting ad. The platform claims that purchase in full, but would it have happened without the ad? Probably yes. That purchase is claimed, but not incremental. Incrementality pulls exactly these two things apart: revenue your ads cause versus revenue your ads merely touch.
For a DTC brand this distinction is not an academic debate. Every euro of budget that goes to claimed but not caused revenue is a euro that does not go to acquiring net-new customers. And growth comes from new customers, not from repackaging demand that already existed.
Branded search campaigns are the second classic example. Someone already knows your brand, types in your name and clicks the ad sitting above your own organic result. The platform reports a beautiful conversion, but chances are that same person would simply have clicked your free result one line lower.
Why is ROAS not proof of incrementality?
ROAS rewards channels that sit close to the purchase. Retargeting, branded search and email flows catch people who had all but decided. They report beautiful numbers as a result, while the revenue they actually add can be limited. Top-of-funnel campaigns, the ones introducing strangers to your brand for the first time, look more expensive on paper, while they create the demand that gets harvested further down the funnel.
Steer purely on ROAS and you systematically reward the harvesting machine while starving the seeding machine. That works for a while, until existing demand runs out and the growth curve flattens. Look at the dashboard then and you will still see great numbers, without understanding why revenue has stalled.
How do you measure incrementality without a data team?
You do not need an econometric model to start. The principle is always the same: deliberately withhold your ads from one group and compare the outcome with the group that saw them. Three approaches that work in practice:
- Geo holdout: temporarily exclude a comparable region from your campaigns and compare its revenue development with the rest of your market. The difference is your incremental effect.
- Segment holdout: exclude part of your retargeting audience and watch how much of that group buys anyway. Whatever comes in by itself was never incremental.
- On and off testing: deliberately lower or pause a campaign and follow total revenue and MER in the weeks after. Crude, but very clarifying when you doubt a large budget.
Important for every variant: define upfront what you measure and how long the test runs, and keep the rest of your marketing as stable as possible. A holdout with a discount campaign cutting through the middle of it mostly measures noise.
Do not expect a black and white outcome either. Almost no channel is fully incremental or fully hollow; you are looking for the ratio. The result of a good test is not a verdict but a correction factor: this much of the claimed revenue turns out to be real, and with that knowledge you reweigh your budgets.
ROAS measures what you may claim. Incrementality measures what you cause.
How does this change your scaling decisions?
Once you think incrementally, your budgets shift. Channels that mostly claim get less, channels that demonstrably cause get more. In practice that almost always means: more weight on top-of-funnel creatives that convert strangers, and a more honest story about what retargeting truly adds. This is exactly why we see creative as the engine behind growth: it is the part of your funnel that creates new demand instead of harvesting existing demand.
It also changes how you look at scaling itself. The question is not whether your average ROAS can absorb a budget raise, but whether the extra euros produce incremental revenue. Brands that have this sharp dare to scale further than their dashboard seems to justify at first glance, because they know which revenue genuinely comes from their spend.
The reverse holds too: sometimes a test shows that a channel you confidently funded mostly redistributes existing demand. That stings for a moment, but it is exactly the knowledge you need to grow from €15 to 20K per month toward €150 to 200K without fooling yourself with your own dashboard.
Conclusion
Measuring incrementality does not need to start big or expensive. One well-designed holdout test often tells you more about your true return than months of staring at dashboards. Start with the channel you doubt most, usually retargeting, and let the outcome sharpen your budget allocation. Curious how to set up a test like this for your brand? Book a call and we will gladly look at your numbers together.