Attribution

    What Is Geo Lift Test?

    A geo lift test is a controlled incrementality experiment where you split your regions into test and control groups, running ads in test regions, pausing them in control, and measure the revenue difference to isolate the true incremental impact of the channel.

    Md Morshed Parvej Patwary
    By
    Updated · Next review

    Design

    You pause Meta ads in 5 comparable cities (control) for 4 weeks, keep them running in 15 similar cities (test). Compare revenue trends. If test cities do 20% more revenue than control (after normalising for baseline), that's your Meta lift. Divide by test-city Meta spend for true incremental ROAS.

    Benchmarks

    • Minimum test length: 3 to 4 weeks (shorter is noise).
    • Minimum spend: ~$30K in the tested channel over the window.
    • Statistical power requires at least 8 to 10 test + control regions.

    Why it matters

    Every attribution model overclaims. Geo lift is the truth check, the one experiment design that doesn't rely on tracking, cookies, or platform self-reporting. Every serious brand runs one at least yearly.

    Common mistakes

    • 1.Testing during holiday or launch periods. Confounds the signal.
    • 2.Using non-comparable regions. If test cities are younger + richer, the 'lift' is demographic, not causal.
    • 3.Testing for 1 week. Signal doesn't stabilise.

    Put Geo Lift Test to work

    FAQs about Geo Lift Test

    Do I need special software for geo lift?

    Not required, but tools like Haus, Recast, or Meta's Lift API make the statistical design easier and more rigorous.

    How often should I run one?

    Annually for major channels, quarterly if you're actively rebalancing budget.