Incrementality is the number of conversions your ads caused that wouldn’t have happened without them, usually measured as the lift of a group that could see the ads over a group that couldn’t. Attribution decides which ad gets credit for a conversion. Incrementality asks whether the conversion needed an ad at all.
Incrementality vs attribution
| Attribution | Incrementality | |
|---|---|---|
| Question | which ad gets credit for this lead? | how many leads happened because of the ads? |
| Method | rules or a model applied to observed clicks and views | an experiment with a control group |
| Sees people who weren’t shown the ad | no | yes, the control group |
| Typical output | conversions per ad, cost per result | incremental conversions, lift, cost per incremental conversion |
| What it costs you | tracking | a holdout that gets no ads while the test runs |
Attribution can’t answer the incrementality question, however good the tracking, because it only ever looks at people who saw or clicked an ad. A view-through lead from someone who was going to fill in the form anyway looks exactly like one the ad persuaded.
How a lift test works
Meta describes its Conversion Lift studies like this: a randomized test group of accounts can see your ads, a control group doesn’t, and Meta compares conversions between the two to determine the increase your campaign generated (Meta for Developers). The arithmetic is simple:
- Split the audience at random, for example 90% test and 10% control.
- Run the campaign. Only the test group can see it.
- Scale the control group’s conversions up to the test group’s size.
- Incremental conversions = test conversions − scaled control conversions.
Example: the test group produces 270 leads and the control group 20. Scaled up nine times, the control stands for 180 leads that would have happened anyway, so the ads caused 90 and the lift is 50%. On €3,000 of spend, that’s €11.11 per lead and €33.33 per incremental lead. Same campaign, a number three times higher.
Meta Conversion Lift: who can run it
- Through the API. Meta’s developer guide opens with a warning that Conversion Lift Measurement is currently limited, and sends you to your Meta representative for access.
- In Ads Manager. Meta’s help center says some Conversion Lift tests display spend minimums or recommended spend amounts when you start creating them (Meta Business Help Center). Check what your own account is offered before you plan a quarter around a test.
- Incremental attribution is something else. Meta also offers an incremental attribution setting, which it measures against its “standard attribution model” (Meta, January 2026). It’s a model Meta runs, not an experiment on your campaign, so it doesn’t replace a holdout.
A lift test needs volume; a CPL you can trust needs only clean data. DropCPL gives you a click-based, first-party CPL per creative and per landing page, at any volume. Start free →
When a small lead gen account can (and can’t) run one
The limit is the number of conversions in the control group. With a 10% holdout, the control sees a tenth of the audience, and a tenth of a small number is a very small number. If the ads did nothing, the 270 leads of the earlier test group would come with about 30 leads in the control group. Here’s what two other outcomes look like:
| Example: 270 leads in the test group | Leads expected in the control | Gap vs “ads did nothing” (30) | Random noise on that count, 95% range | Readable? |
|---|---|---|---|---|
| Half the leads caused by the ads | 15 | 15 | ±8 | yes |
| One lead in ten caused by the ads | 27 | 3 | ±10 | no |
An account of this size can see that its ads double its leads. It can’t tell a lift of around 10% from zero, and that’s usually the answer people run the test for. Volume, more than tooling, decides whether a lift test is worth paying for.
If you’re below that line:
- Decide on click-based, first-party numbers and treat view-through and modelled credit as unproven. Real CPL vs platform CPL makes the full case.
- If you test anyway, test big. Pause a whole campaign in one region for a few weeks and compare it with similar regions. It’s cruder and exposed to seasonality, but it asks the right question.
- Don’t confuse it with an A/B test. An A/B test compares two versions of an ad or a landing page. A lift test compares ads with no ads.
Related terms
- View-through attribution: the credit incrementality questions most
- Attribution window: how long attribution lets a click or view claim a conversion
- Real CPL: the number to decide on when you can’t test
- A/B testing: comparing variants rather than ads against no ads