# What is a conversion lift study?

> A conversion lift study splits people into an ad group and a hold-out group to show how many extra sales your ads caused. Meta's method, an example and limits.
- **Author**: Marxx
- **Published**: 2026-11-17
- **Category**: Measurement
- **URL**: https://marxx.ai/library-glossary/conversion-lift-study

---

A conversion lift study shows how many extra sales your ads caused. It splits people into a test group that can see your ads and a control group that cannot, then compares their sales. The gap is the lift ([Meta Help Center](https://www.facebook.com/business/help/453968412201452), checked 2026-10-20).

## How the maths works

Meta works out incremental conversions from the difference between the test group and the control group. It scales the control to match the size of the test group first. It then gives a range of likely values, using a 90% credible interval ([Meta lift glossary](https://www.facebook.com/business/help/1092662031214127)).

- Incremental conversions = test group conversions − scaled control conversions
- Cost per incremental conversion = amount spent / incremental conversions

## A worked example

Say you spend INR 2,00,000 on a lift study. The numbers are made up.

| Group | Purchases |
|---|---|
| Test group (could see ads) | 5,200 |
| Control group, scaled to the same size | 4,700 |
| **Incremental** | **500** |

- Cost per incremental purchase = 2,00,000 / 500 = **INR 400**
- Ads Manager credited the ads with 1,000 purchases, or INR 200 each.

So about half of the credited sales would have happened anyway. Your true cost per extra sale is double what Ads Manager says. If your product leaves INR 600 after costs, you still make money. If it leaves INR 300, you do not.

## Why use one

Ads Manager credits sales that follow an ad. It cannot tell which of them the ad caused. A lift study can. Meta calls incrementality experiments, like Conversion Lift, the best way to answer what the ads caused ([Meta, 3 March 2026](https://www.facebook.com/business/news/click-attribution)). See [incrementality](/glossary/incrementality).

## What to know before you run one

- Meta lists lift tests among its Facebook-managed tests. You set them up with your Meta account representative, who knows the minimums.
- Meta says that Conversion Lift tests which began measuring standard web events on or after 1 October 2021 need the Conversions API. See the [Conversions API](/glossary/conversions-api).
- Results are estimates with a range, not exact counts.
- A smaller budget or a short test may give a range too wide to act on.

If you cannot run a Meta study, try a [holdout test](/glossary/holdout-test) or a [geo lift test](/glossary/geo-lift-test).

## FAQ

### Is a conversion lift study the same as an A/B test?

No. An A/B test compares two ads. A lift study compares people who could see your ads against people who could not.

### What is a good lift?

There is no standard number. Compare the cost per incremental sale with your profit per sale.

### How long does a study take?

It depends on your budget and sales volume. Ask your Meta representative for the minimums.

---
- [More Measurement articles](https://marxx.ai/library-glossary/category/measurement)
- [All articles](https://marxx.ai/library-glossary)