Holdout and GeoLift
Both run geo holdout tests. Both use rigorous methods. The difference is who does the work of designing, executing and living with the test. Facts about GeoLift are from its public documentation (checked July 2026) and its pricing page (re-checked 28 September 2026: unchanged).
What GeoLift is
GeoLift is Recast's self-serve geo experiment tool. It is inexpensive, statistically serious and flexible: go-dark, spend-down and spend-up tests, online and offline channels, any geographic units you can supply, simulation-based power analysis. A measurement practitioner who already knows what test they want can run it well.
The work it leaves with you is the point. You choose the test type, budget, duration, power target, number of regions, exclusions, cooldown and how to aggregate geography. You make the media change in the ad platform by hand and take it back out by hand. You upload the outcome data and download the analysis. If you have a measurement team, that is a reasonable division of labour.
What Holdout is
Holdout is built for the performance marketer who needs the answer rather than the tooling. It reads your orders, designs the test, writes the exclusions to the ad account, watches that they hold, removes them on the end date, measures the result and then applies the measured rate to every campaign in the channel so the dashboard you already use reads true. Priced per test, no contract, no sales call.
The price of that is scope. Holdout runs one kind of test: switching a channel off in a set of UK regions. It does not run spend-up tests, offline channels or non-UK geography today.
Side by side
GeoLift
Self-serve for a measurement practitioner
You design it, you make the media change, you upload and download.
Holdout
Self-serve for a performance marketer
It designs, writes the pause, watches it, measures, and keeps the dashboard corrected.
| GeoLift | Holdout | |
|---|---|---|
| Price | $100 a month or $1,000 a year per user seat, after a six-month free trial | £349 per test. £299 a month for ongoing reporting |
| Who runs the test | You: test type, budget, duration, power, region count, exclusions, cooldown, aggregation | The design tool: one recommended plan from your orders, with the precision it will deliver |
| UK geography | Any units you supply; automatic mapping is US postcodes to commuting zones | UK postcodes mapped to the regions Google Ads and Meta can actually target |
| Pausing the channel | You make the media change in the ad platform yourself | Writes the location exclusions to Google Ads, checks they hold, removes them on the end date |
| Test types | Go dark, spend down, spend up; online and offline channels | Go dark on a channel you can pause by region |
| The result | A point-in-time analysis you download; exports into Recast MMM | A measured rate that keeps correcting every campaign in the channel until you re-test |
| Data source | Upload | Shopify connection or an orders export as downloaded |
Pick GeoLift if
- You have someone who designs experiments for a living and wants full control of the design.
- You need spend-up or spend-down tests, offline channels, or geography outside the UK.
- You already run Recast's marketing mix model and want the experiments feeding it.
Pick Holdout if
- You run UK paid media and want to know what a channel really drives without becoming a statistician.
- You want the pause applied and removed in the ad account, and checked while it runs, rather than done by hand.
- You want the result to keep working after the test: a rate that corrects every campaign until the next test.
- You want to pay for one test and decide afterwards.
What both will tell you
The method is not anyone's secret. Calibrating what a platform claims against what a holdout measures is how every serious measurement team works, and either tool run properly will get you to the same kind of number. The choice is about who does the running. If you are not sure a test can work for your volume at all, the readiness checker answers that first, and it applies whichever tool you pick.
See what a test would look like on your sales.
Design is free. Nothing changes in your ad account until you say so.
Design a test