Causal measurement · Growth Planning Studio
Make sure the test is worth running, before you run it.
A geo-holdout experiment simulator and planner. Size the test before you spend, then read incremental lift, its interval and iROAS when it runs.
Live tool · one self-contained file · yours to keep
The problem it solves
Platform ROAS counts conversions the ads may not have caused.
The clean read is a geo-holdout: keep some markets dark, run media in the rest, and measure the gap against what would have happened anyway.
The catch is that an underpowered test comes back inconclusive. Weeks gone, and holdout revenue given up for nothing. This tool sizes the test first: how many markets, how long, how big a holdout, to detect the lift that would actually change a decision, at the confidence you want.
Then it reads the result: incremental lift, its interval, and iROAS, the true revenue per dollar of spend. The finance number, not the platform's.
How it's different
Most tools sell you the test. This one makes sure it's worth running first.
Simulate a test to see how lift, significance and power behave. Switch to Plan to size markets, duration and holdout against a target effect and power before you commit.
It works in both frequentist and Bayesian modes, and the Bayesian mode can fold in a prior from a previous test or your MMM, so a smaller test can still conclude. Every input is explained in plain language, and a built-in note shows exactly how the numbers are computed.
In your hands
What you get.
- Simulate modeInject a known lift into synthetic markets and watch how the estimate, significance and power respond. The safest way to build intuition before a real test.
- Plan modeSize the test to detect your minimum effect at 80 to 90% power, with the smallest detectable effect at each market count.
- Frequentist and BayesianRead the lift either way. The Bayesian mode pulls toward a prior from a past test or your MMM.
- The decision numbersTreatment versus counterfactual, incremental lift with its interval, and iROAS.
And the rest
One self-contained file. No login. Works in any browser, exports any chart or table to CSV or image. A walkthrough of the findings and a refreshed build when new data lands, all included.
- No login, no seat to learn
- Exports to CSV or image
- Walkthrough of the findings included
- Refreshed build when new data lands
Alternatives
What it beats.
Open-source geo-lift
Meta GeoLift and Google CausalImpact are the gold-standard method, but R-only: you need an analytics team to run it.
Managed services
They run tests for you, but as a subscription, with the design inside their black box.
Platform lift tools
Quick and free, but single-channel, and they need outside validation.
Not sure if this is the one?
Eight quick questions. Within 24 hours we'll come back with the solutions worth considering and the approximate budget.