Attribution and Incrementality
Know what actually drives growth, not just what the platforms claim.
For: Multi-channel B2C businesses
Google claims 100 conversions. Meta claims 80. TikTok claims 40. Your CRM shows 90 total. Every platform is grading its own homework, and when you add up the dashboards you are counting close to 1.6 sales for every one that actually happened. Since iOS and cookie loss, the platforms model the conversions they can’t see, so the numbers are inflated in ways you can’t audit from the inside.
Attribution assigns credit. Incrementality proves cause. The only way to know what your ads actually drove, rather than what would have happened anyway, is to hold a channel back and measure the difference. We design geo holdout and lift tests, reconcile what the platforms report against what really landed, and deduplicate the double-counted conversions into a single source of truth.
Then the test has to change something. We turn the results into a reallocation plan, moving budget off the channels that were barely incremental (usually brand search and retargeting) and into the ones that actually grow revenue. This needs scale to be worth it. If a single channel is spending under roughly 40.000€ a month, we will tell you to run a free platform lift test instead.
At GetYourGuide, Ilja ran blackout A/B tests across the display channels and validated incrementality at a multiplier of around 2.8x for Smart Display, the kind of evidence that moves a budget with confidence instead of a hunch.
Who this is for
B2C teams spending across three or more channels who can't say which one actually drove the revenue.
Common questions
Questions, before we talk.
Attribution assigns credit for a sale. Incrementality proves the ad caused it. A conversion can be credited to a channel that would have converted anyway; only a holdout test tells you which.
Partly. Platform-reported ROAS is usually inflated, worst on brand search and retargeting. We reconcile what the platforms claim against what actually landed.
We hold a channel back in matched regions and measure the sales difference versus regions where it kept running. The gap is the true incremental effect.
This needs scale. If a single channel spends under roughly 40.000€ a month, we'll tell you to run a free platform lift test instead of paying for a full study.
Only at real scale: 1M€+ annual spend and two-plus years of clean data. Below that the model is noise, and we won't sell it to you.
Didn't find your question?
Send Ilja a message →(—) Next step
Sure your Meta ROAS is real?
Book a 15-minute intro call. We'll talk through your channels and where the numbers might be inflated, and whether a holdout test is worth it for you yet. No pitch.