Measurement · 02 Demo · synthetic data
Geo test readout
Did the spend cause the sales?
Test markets against a pre-period-scaled control. A lift estimate with an uncertainty band and a keep, cut, or keep-testing call against break-even.
Geo test readout
Four test markets get $8.0k a week of extra spend. Six control markets don't.
Market noise
Break-even: 2.50× revenue per test dollar
The call
Keep
The whole 95% band clears break-even. The spend pays for itself in these markets.
Measured lift
+11.8%
95%: +7.5% to +16.1%
Incremental revenue
$208k
$132k to $284k
Return per test dollar
4.34×
2.76× to 5.92× · break-even 2.50×
Pre-period fit error: 4.6% of weekly sales. Control markets are scaled to match the test markets over the 12 pre-period weeks; the band comes from how far the two drifted apart before launch. The true effect slider exists only because this is a simulator.
Synthetic numbers generated in your browser. Not client or employer data.
How to read it
What the demo is showing
Before the test, control markets are scaled so they track the test markets. After launch, that scaled control is the estimate of what test markets would have sold without the spend.
The band comes from how well the control tracked during the pre-period. A noisy pre-period means a wide band, and a wide band often means the honest answer is to keep testing.
The call compares the band, not the point estimate, against break-even: keep when the whole band clears it, cut when the whole band misses it.
The shift from MTA to macro-measurementHow I'd approach an MMMHow I'd design a geo testNext demo: Experiment power calculatorAll demos