Preset
Standard
95% confidence80% power10% lift goal
Test setup
Stat assumptions
Traffic model
Ad Test Significance Calculator
Plan PPC test sample size, expected duration, and likely finish windows in one place.
Ad Test Significance Calculator
Outputs
Run the calculator to estimate the sample size and likely finish window.
Start with the assumptions on the left. Once you run the model, the tool will tell you how much volume each variant is likely to need and how long that should take at the budget you entered.
If you want the deeper read after that, you can open the timing explorer for faster, typical, and slower finish scenarios.
Needed per variant
How much volume each arm of the test needs before you should expect a stable read.
Total sample
The combined observation target across control and variant.
Expected finish
A rough completion window based on the CPM and daily budget you enter.
How This Calculator Works
Use the planner cards for headline numbers, then use the Monte Carlo view to judge fast, typical, and slower finish outcomes.
How to use it
- 1.
Set the baseline rate and the smallest lift worth detecting.
- 2.
Match alpha, power, CPM, and budget to the way you actually run tests.
- 3.
Use the top cards for planning and the Monte Carlo explorer for finish timing.
Glossary
- Confidence
- How strict the significance threshold is. Higher confidence usually needs more data.
- Power
- How likely the test is to detect the chosen lift if that lift is real.
- Lift goal
- The minimum relative improvement you want the calculator to plan for.
- Bayesian win rate
- A modeled read of how likely variant B is to beat the control under the assumed rates.
Method
- Sample size planning uses a two-proportion significance test.
- Expected finish timing is estimated from CPM and daily budget via daily impression volume.
- Monte Carlo paths simulate cumulative control and variant performance over time.
- Sequential checks are flagged because repeated peeking can inflate false positives.
Frequently Asked Questions
Common questions about sample-size planning and simulated test duration.