Tool
Incrementality Calculator
Plug in your test and control group results to measure incremental lift, incremental CAC, and statistical significance, instantly.
New to incrementality? Read the incrementality definition
Population = unique users (or sessions, or geos) in each cell. Required for a two-proportion z-test.
Your numbers
You can defend this lift to a CFO. Consider a second test window before permanently reallocating budget.
Rule of thumb: if incremental CAC is dramatically higher than your platform-reported CAC, the channel is over-credited and likely needs a budget cut.
Target lift benchmarks by channel (2026)
Directional bands from live geo-lift and holdout tests I have run across Bangladesh and international accounts. Use these to decide whether to keep, cut, or restructure a channel.
| Channel | Healthy lift | Suspect zone | Action |
|---|---|---|---|
| Branded Search | 10-25% | Below 5% | Cap budget, do not scale |
| Non-brand Search | 40-70% | Below 20% | Rebuild keyword tiers |
| Meta prospecting (broad) | 25-60% | Below 10% | Refresh creative, retest |
| Meta retargeting | 5-15% | Below 3% | Cut, most is cannibalisation |
| Performance Max | 20-45% | Below 10% | Exclude brand, audit feed |
| YouTube / Demand Gen | 8-20% | Below 3% | Reserve for warm audiences |
How to Calculate Incremental Lift
Incrementality testing isolates the conversions your ads caused from the conversions that would have happened anyway. You hold out a matched control group from all advertising, then compare results.
Incremental Conversions = Test Conversions − Control Conversions Lift % = (Test − Control) ÷ Control Incremental CAC = (Test Spend − Control Spend) ÷ Incremental Conversions Example: Test: 520 conv from $8,000 spend Control: 410 conv from $0 spend Incremental = 520 − 410 = 110 conversions Lift % = 110 / 410 = +26.8% iCAC = $8,000 / 110 = $72.73
When to Run an Incrementality Test
- Scaling budgets beyond $50k/mo, small reporting errors become big dollar leaks.
- Launching new offline or top-of-funnel channels where click attribution doesn't apply.
- When platform CPA drastically disconnects from backend CAC.
What is an incrementality test?
An incrementality test (also called a lift study or geo-holdout) measures the true causal impact of advertising by comparing a group exposed to ads (test) with a matched group that was not (control). The difference between the two is the incremental conversions your ads actually caused, separating real ad-driven sales from sales that would have happened anyway.
How do you calculate incremental lift?
Incremental Lift % = (Test Conversion Rate − Control Conversion Rate) ÷ Control Conversion Rate. Incremental Conversions = Test Conversions − (Control Rate × Test Population). Incremental CAC = Test Spend ÷ Incremental Conversions.
When is a lift test statistically significant?
At a 95% confidence level, a two-proportion z-test result with |z| ≥ 1.96 is considered significant. If your sample sizes are small or the lift is marginal, you usually need more traffic or a longer test window.
When should you run an incrementality test?
Run incrementality tests on always-on channels (branded search, retargeting, Performance Max, broad prospecting on Meta) where platform-reported ROAS overstates true contribution. They are especially valuable post-iOS 14 as click-based attribution has become noisier.
Not sure if your ads are actually driving growth?
Book a free growth gap audit, we'll design a geo-lift or holdout test that gives you a defensible answer.
Email me this lift result
Send me your test and control numbers and I'll come back with a short note on whether the lift is real and what to do next.