Paid Media

Incremental Lift Calculator

Attribution tells you which ad got clicked. A holdout tells you what would have happened anyway. This calculator compares an exposed group against a control group so you can see the conversions your spend actually created.

Your numbers

Result

Conversion rate lift50.0%
Incremental conversions600
Incremental CPA$41.67
Test group conversion rate3.00%

FormulaLift % = (test conversion rate − control conversion rate) ÷ control conversion rate × 100. Incremental conversions = test conversions − (control rate × test audience)

Worked example

1,800 conversions from 60,000 exposed people is a 3 percent rate, against 1,200 from 60,000 held out, a 2 percent rate. That is a 50 percent lift and 600 genuinely incremental conversions. At $25,000 of spend the incremental CPA is $41.67, while the platform would happily report a CPA of $13.89 by taking credit for all 1,800.

What to watch for

The gap between reported CPA and incremental CPA is the most uncomfortable number in performance marketing, and on retargeting campaigns it is often enormous. Running one honest holdout usually reshapes a budget more than a year of optimization does.

Small holdouts produce unstable results. If your control group is a few thousand people with a handful of conversions, the lift figure will swing wildly, so size the test for the conversion volume you actually have before you trust the output.

Keep the holdout genuinely excluded across every channel you can control. A control group that still receives your email and organic social is measuring paid media against paid media plus everything else, and the answer will be wrong in a direction you cannot predict.

Frequently asked questions

How big should my holdout be?

Big enough that the control group produces a stable conversion rate, which usually means at least a few hundred conversions. As a starting point, a 10 percent holdout on a high-volume campaign is common, and smaller campaigns need a proportionally larger share.

Why is incremental lift lower than attributed conversions?

Because some of the people who converted after seeing your ad were going to convert anyway. Attribution assigns credit for the click, incrementality measures the change in behaviour, and only the second one is a reason to keep spending.

How often should I run lift tests?

At least once per major channel per year, and any time you are about to make a large budget change. They cost real conversions in the short run and repay it by preventing much larger misallocations.

Read next

You might also like

Ranked by how closely each page overlaps with this one, using a similarity model over the whole library.

More Paid Media calculators

This calculator runs entirely in your browser. Nothing you enter is sent to a server, logged, or stored. Figures are for planning and do not constitute financial advice. See disclosures.