Every marketer talks about incrementality. Few can define it.
Here's the definition: Incrementality is the revenue you generated because you ran the campaign, not the revenue that happened to occur after you ran the campaign.
The distinction matters more than anything else in marketing measurement.
The problem with correlation
Let's say you run a retargeting campaign and track 5,000 conversions. Your analytics platform reports a 4× ROAS. Success, right?
Not necessarily. The question is: how many of those 5,000 people would have converted anyway, even without seeing your ad?
If the answer is 4,500, your campaign only drove 500 incremental conversions. Your true ROAS isn't 4× — it's 0.4×. You're losing money.
Retargeting campaigns are the classic example. They target people who already visited your site and showed purchase intent. Many of those people were already coming back to buy. The ad didn't cause the conversion — it just happened to appear before one.
How to measure incrementality
The gold standard is a holdout test:
- Split your audience into two random groups: treatment (sees your ad) and control (doesn't see your ad)
- Run the campaign for the treatment group only
- Measure conversions in both groups
- The difference is your incremental lift
Example: Treatment group converts at 3.2%. Control group converts at 2.8%. Your ad drove a 0.4 percentage point lift, or 14% incremental conversions. That's your real impact.
Why most platforms don't show incrementality
Meta, Google, and TikTok report attributed conversions — the number of people who saw your ad and then converted. They don't subtract the people who would have converted anyway.
Why? Because holdout tests reduce spend. If you withhold ads from 50% of your audience, the platform earns less. It's not in their interest to make incrementality testing easy.
But it's in your interest. If you're spending $500k/month and 40% of it is non-incremental, you're burning $200k.
The bottom line
Attribution tells you who converted. Incrementality tells you why.
If you're optimizing for attributed conversions, you're chasing credit for things that would have happened anyway. If you're optimizing for incremental conversions, you're maximizing real growth.
Run a holdout test. The results might hurt. But they'll save you from wasting six figures on campaigns that don't work.