Incrementality asks the question every advertiser should: which of these sales would not have happened without the ads? Platform ROAS counts sales that followed an ad. Incrementality counts only the ones the ads caused. The gap between the two can be large, and it decides whether scaling is actually profitable.
Why ROAS can overstate impact
- People who were going to buy anyway. Loyal customers, people who searched for your brand and shoppers already mid-checkout often see an ad before buying, and attribution gives the ad the credit.
- View and engagement credit. A 1-day view window can credit an ad someone scrolled straight past.
- Warm audiences. Recent visitors buy at high rates with or without ads, so retargeting usually looks better in attribution than in lift tests.
The reverse can happen too. Ads can drive sales that attribution misses, such as purchases made days later on another device, which is why you need a test rather than a guess.
How incrementality is measured
Conversion lift (holdout) tests. Meta randomly splits an audience into a test group that can see your ads and a control group that can't, then compares purchases between them. Meta offers these through its Experiments tools for eligible accounts, so check what yours can access.
Geo holdout tests. Pause (or start) ads in a set of comparable cities or regions and compare sales with regions left unchanged. Meta's open-source GeoLift package is one way to design and read these tests.
Before-and-after checks. Pausing a campaign and watching total store sales is crude, because seasonality and other channels muddy the picture, but it beats never checking.
Worked example
Example: a brand spends ₹20,000 (about $206) during a lift test. Ads Manager reports ₹80,000 of attributed revenue, a ROAS of 4.0. The lift test shows the test group placed 240 orders against 200 from an equal-sized holdout: 40 incremental orders at ₹1,000 each, or ₹40,000. Incremental ROAS is 2.0. If the brand's break-even ROAS is 1.8, the campaign still makes money. If it's 2.5, it loses money, despite the 4.0 on the dashboard.
How to use the result
Run a lift test before big budget decisions rather than every week. Use the result as a correction factor: if platform ROAS runs at roughly double your incremental ROAS, your platform target needs to be roughly double your break-even. Between tests, track MER, total revenue divided by total ad spend, as a sense check. Our guide to calculating break-even ROAS shows how to set the underlying target.
Frequently asked questions
What is incrementality in marketing?
Incrementality is the extra sales, sign-ups or other results caused by a campaign, compared with what would have happened without it. It's measured by comparing a group exposed to ads with a similar group that wasn't. It's a stricter test than attribution, which credits any sale that followed an ad.
How do I run a conversion lift test on Meta?
Meta offers conversion lift studies through its Experiments tools for eligible advertisers. They randomly hold out part of your audience from seeing ads and compare purchases. Availability and minimum requirements vary, so check the Experiments section of Ads Manager. You need reliable purchase tracking and enough volume for a readable result.
What is incremental ROAS (iROAS)?
Incremental ROAS is the revenue your ads caused divided by what you spent. It's often lower than platform ROAS because it excludes sales that would have happened anyway, though it can be higher when ads drive sales that attribution misses. Use it to decide whether a campaign truly earns its budget.
