Know what your marketing actually caused.
A controlled test that separates sales your ads created from sales you’d have made anyway.
Included in every plan, free and up. No credit card.
Trusted by growth teams who left enterprise measurement vendors
Every platform says it drove the sale.
Credit isn’t cause. The most direct way to know what your ads caused is to hold them out in matched markets and measure the difference.
The test is won in the design.
Matched control markets, with promos and price changes in the design instead of the footnotes. If the markets didn’t track before the test, no model rescues it afterwards.
Models compete. Best fit wins.
Selection happens on pre-period fit, before the treatment effect is ever looked at.
Illustrative. Best-fitting candidate on the pre-period criteria, chosen blind to the result.
Know what happened here.
See the estimated impact under the spend, timing, and conditions you tested.
Now turn the result into a curve.
Your holdout anchors the model, so Stella can estimate what happens beyond the spend level you tested.
Illustrative.
With Stella, you know.
What a channel actually caused.
The revenue that only happened because the ads ran.
Whether they would have bought anyway.
The conversions your platforms are counting twice.
Your incremental ROAS.
Not the platform number. The one that reconciles.
Whether branded search is incremental.
The test most brands avoid running.
What happens if you turn it off.
Before you find out the expensive way.
Whether the result is solid enough to act on.
The range and the significance, not a lone point estimate.
Ask Stella about holdouts.
Where to run one, how much spend it takes to detect lift, or what a study you already ran really showed.
Hear it from the brands we run measurement for.
Incrementality questions, answered.
How does Stella pick test and control locations?
Location selection is where most holdouts are won or lost. Stella analyzes your pre-period history and selects control markets that have moved together with your test markets, so the comparison is valid before the test even starts. It works even for dense, multi-location footprints, using a weighted synthetic control.
How does Stella choose which model to trust?
Several candidate models compete on pre-period fit, how closely each one tracked the test market before anything changed. The best-fitting candidate on those criteria is selected before the treatment effect is looked at, so nobody can shop for the result they wanted. You see the pre-period R² and MAPE alongside the estimate and its interval.
How long does a holdout take?
A geo holdout typically runs 20 to 28 days, longer for high ad-stock channels. If you want a usable read sooner, Always-On Incrementality (Professional and up) produces daily directional iROAS while the holdout incubates the rigorous number.
What can I do on the free tier?
Run a real holdout study, no credit card. Free includes one holdout and one media mix model a year, on data you upload. Already ran a holdout elsewhere? Drop it in and see what our location selection would have recommended before you upgrade anything.
What plan includes incrementality testing?
Every plan, free and up. Always-On Incrementality, which keeps the number current between tests, starts at Professional ($3,000/mo).
Know what your marketing actually caused.
Run a real holdout on your own numbers, free. Or have a founder design the first one with you.