Incrementality

What is incrementality?

Incrementality measures whether your ad campaign actually caused conversions — or whether those users would have converted anyway without ever seeing your ad.

It’s the difference between two very different claims:

• “We got 1,000 conversions during the campaign.” True but potentially meaningless. How many of those people would have shown up on their own?
• “We got 1,000 conversions, and 800 of them wouldn’t have happened without the ads.” Now you actually know what the ad spend bought you.

That second number — the conversions genuinely caused by the campaign — is the incremental lift. Everything else is noise that attribution alone can’t untangle. A campaign can look great on paper (lots of conversions, strong attribution) and still be adding almost nothing real if most of those users were going to convert anyway.

Incrementality testing is how you separate the two. It’s the only honest way to know whether an ad campaign is working.

Why nobody in web3 does this

Incrementality testing is genuinely hard. Even Google and Meta — with every behavioural data point imaginable and armies of statisticians — treat it as an advanced study that most advertisers never run properly. It requires careful experimental design, large sample sizes, and the ability to construct credible control groups.

In web3, nobody is doing this. Every other web3 ad network reports on attribution alone — conversions that happened during the campaign, often without accounting for users who would have converted regardless. That number is fine as a vanity metric but doesn’t tell you whether your spend is doing any real work.

Specify runs an incrementality study on every campaign, starting with the test campaign.

How we do it

Standard incrementality testing uses a control group — a randomly selected subset of your target audience that deliberately sees no ad, so their conversion rate can be compared against users who did see the ad. The gap between the two is your incremental lift.

We do it differently.

Rather than running a traditional control group, we build a lookalike audience drawn from onchain behavioural data. This is only possible in web3, where the relevant behavioural history is already public. The lookalike audience:

• Is not active in the advertised product at the time of the study
• Matches the exact same targeting criteria as the treatment group
• Is filtered to match the treatment group’s latest transaction recency, so activity level distribution are matching
• Has our wallet-linking technology applied the same way — so if a control user converts across different wallets, we catch it too

The only meaningful difference between control and treatment is whether they saw the ad.

From there we measure the baseline conversion rate of the lookalike audience and compare it to the treatment group’s conversion rate. The ratio between the two is the incremental lift.

What the results tend to look like

In web2 advertising, a 0.5x incremental lift is generally considered a strong campaign result — and even that varies heavily by industry and campaign type. Many campaigns deliver much less.

On Specify, we frequently see 4-6x incremental lift. Six times more conversions in the treatment group than the lookalike baseline. That’s the difference onchain behavioural targeting makes when it’s pointed at the right users.

Because of how we construct the lookalike audience, we also get a large sample size on the control side — which gives us statistical significance typically greater than 99%. You’re not looking at a suggestive pattern; you’re looking at a result that would be vanishingly unlikely to arise by chance.

Where you see this

Every test campaign report includes a full incrementality analysis — baseline conversion rate, treatment conversion rate, lift ratio, and statistical significance. We walk through it with you on the post-campaign call alongside the volume and targeting analysis.

For ongoing campaigns, incrementality remains part of the reporting cadence so you always know not just how many conversions you bought, but how many of them your ad spend actually caused.

Want to go deeper?

We cover incrementality in considerably more detail — along with the rest of the web3 performance marketing stack — in our Web3 Advertising Course .

The course is free for Specify clients. If you want to understand the statistical foundations behind what we measure, how to interpret incrementality results critically, and how to apply the same thinking to the rest of your marketing, it’s the fastest way in.