How to test if your Meta ads are incremental, step by step

Usually with Meta's own Conversion Lift test. Check that your account qualifies and write down the return your margin needs. Then create the test in Experiments and leave the campaigns alone until it ends. Judge the ROAS lift against that bar, not against Ads Manager. If you don't qualify, run a regional holdout instead.

A Conversion Lift test is Meta's version of a holdout. Meta splits the campaign's intended audience at random into a test group and a holdout group. People stay in their group for the whole test, and the holdout group is kept away from your ads. The gap in conversions between the two groups is the lift.

This route suits a store whose biggest question mark is Meta. If your doubt sits with Google, the go-dark route in How to run an incrementality test, step by step fits better. Every menu path below comes from Meta's, Shopify's or Google's own help pages.

Step by step

1. Check that you qualify. As a guide, Meta wants a campaign that started in the past year with $5,000 USD or more in spend. That campaign also needs at least 500 conversions under a 1-day click, 7-day click or 1-day view setting. Self-serve tests need a clean signal too, such as Conversions API events with an event match quality score above 5. Path: Events Manager > your data source > Purchase > View details.
2. Write down the bar first. Divide 1 by your gross margin to get the return each euro of ads needs to pay for itself. Shopify works out gross margin as net sales minus cost, divided by net sales. It does so only for products that had a cost recorded when they sold. Path: Shopify admin > Analytics > Reports > Category filter > Profit Margin > Gross profit by product.
3. Decide what the test should answer. Start broad: Meta suggests a first test at account level, looking at the overall effect of your advertising. Test a single campaign later, once you know the channel as a whole earns its keep. Path: Experiments > Conversion Lift.
4. Set the dates from your buying time. Meta does not recommend using the post-test conversion window for calculating results. So the test itself has to outlast your slow buyers. Read Days to key event with Path length set to greater than 1 touchpoint. Path: GA4 > Advertising > Key events dropdown > Key event attribution paths.
5. Create the test and read the spend guidance. Meta's Experiments tool walks you through a Conversion Lift test step by step. Some tests display spend minimums or recommended spend amounts as you create them. If the guidance beats your plan, run longer or test the whole account. Path: Experiments > Conversion Lift > on-screen steps.
6. Leave the campaigns alone. Meta treats any change to targeting, creative or the optimization event as a significant edit, which restarts learning. Expect your CPM to move: Meta notes lift tests can affect it, because the control group sees no ads. Path: Ads Manager > Ad sets > Columns > add Last significant edit.
7. Read the result in Experiments. Results can appear once at least 100 conversion events are observed, but Meta recommends waiting until the test has finished. Then read the conversion lift, its confidence percentage, the sales lift and the ROAS lift. Path: Experiments > Learn > your test > View report > Results.
8. Judge the ROAS lift against your bar. Meta reports ROAS lift as a multiple of revenue generated for each dollar spent on your Meta ads. Above your break-even, the tested spend pays for itself. Below it, the spend loses margin even though it caused sales. Path: Experiments > View report > Results > Other metrics.

A worked example

Meta's help page gives the core sum. If 5,000 events are observed in the test group and 2,000 in the control group, the conversion lift is 3,000 events. For single-cell holdout tests, the control group is 10 percent of the size of the test group. Meta scales it up to the test group's size before it compares.

Now the money, with round numbers invented for the example.

The bar comes from your margin. The Break-even sheet of one store's export does the sum for a 40% margin: 1 divided by 0.40 is 2.5x. For illustration, a 3.0x lift clears that bar. If your margin is 40%, €120,000 of extra sales carries €48,000 of gross profit against €40,000 of spend. For illustration, that leaves €8,000 ahead before shipping and fees.

For example, Ads Manager showed 5.0x for the same weeks while the lift test showed 3.0x. Don't read the gap as an exact overcount. Meta says lift results are not meant to be compared with Ads Manager, which credits conversions in 1, 7 or 28 day windows. Read the gap as a warning about which number to budget on.

The gap can run the other way, too. In one store's Channels sheet, Direct holds 57.7% of revenue in all three views. If your buyers see a Meta ad and come back days later by typing your address, Ads Manager may not credit the ad. The lift test still counts the purchase, so its result can beat the dashboard.

What to check when the report looks wrong

• Both groups performed about the same. Meta shows this when the difference wasn't conclusive. Wait until the test has finished, because the outcome can still change. If it holds, the effect was too small to see at this size, which is not the same as zero.
• There is no result for Purchase. Meta may show nothing for an event if the test hasn't run long or the groups barely differ. Check the other events in the test, such as Add to cart, before you write it off.
• Confidence sits below 90 percent. Meta treats 90 percent or higher as a statistically reliable result for lift tests. Below that, read the lift as a direction and plan a longer or larger test.
• Meta's purchases don't match your orders. The lift only counts conversions Meta can match to people in its groups, so missing events shrink both sides. Compare a week of Purchase events with the orders in Shopify's Total sales over time. Meta advises aiming for a 75% event coverage ratio of Conversions API to Meta Pixel events.
• Someone edited a campaign mid-test. Check the Last significant edit column for dates inside the test. Note them next to the result, and trust the result less.

What to do this week

1. Check your signal in Events Manager. Go to Events Manager, select your data source, find Purchase and click View details. Pass: Conversions API events arrive next to the pixel's, with event match quality above 5. Fail: pixel only, or a low score, so fix the Conversions API setup before you book a test. If you mainly sell in the EU, this matters twice: Meta says its other signal route is unavailable there.
2. Find a campaign that qualifies. In Ads Manager, set the date range to the past year and sort campaigns by Amount spent. Pass: one campaign clears Meta's spend and conversion guide from step 1. Fail: none does, so run a regional holdout, step by step instead.
3. Put your bar in writing. In Shopify, open Analytics > Reports, filter by Profit Margin and open Gross profit by product for last quarter. Pass: your best sellers all have a cost recorded, so 1 divided by your margin is a bar you can defend. Fail: some don't, so add Cost per item under Products first.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Sources, 1 October 2026: About lift and holdouts in Facebook advertising tests (Meta Business Help Center); About Conversion Lift (Meta Business Help Center); View server event details in Meta Events Manager (Meta Business Help Center); Profit reports (Shopify Help Center); Best practices to get started with Experiments (Meta Business Help Center); Key events attribution paths report (Google Analytics Help); Significant edits and learning phase (Meta Business Help Center); About the learning phase (Meta Business Help Center); About Experiments (Meta Business Help Center); View and understand holdout test results across Meta technologies (Meta Business Help Center); Differences between Conversion Lift test results and other reporting tools (Meta Business Help Center); Similar performance between test and holdout groups in a test (Meta Business Help Center); About confidence in your tests and experiments (Meta Business Help Center); Sales reports (Shopify Help Center).

Related answers

• All Incrementality questions
• How do I test if my ads are incremental?
• Meta Conversion Lift explained: run it and trust it
• Lift test, in the glossary
• Incrementality testing, explained