Multivariate Test Duration Calculator

Work out multivariate test duration instantly with clear inputs, formula shown and shareable results.

About Multivariate Test Duration Calculator

Sample size scales with baseline variance and inversely with the square of the effect you want to detect, then multiplies by the number of variants. That is why adding variants to a low-traffic test makes it practically undetectable.

Formula

Sample size

Per variant ≈ 2 × 7.849 × p(1-p) / (p × MDE)² at 95% confidence and 80% power

Worked example

With the default values below, Multivariate Test Duration Calculator returns Days Of Traffic Required of 25.6. Change any input above and every figure here is recalculated in your browser.

Inputs used

Number Of Variantsaccepts 2–20: 4
Daily Visitorsaccepts 1–10,000,000: 3,000
Baseline Conversion Rateaccepts 0.05–100 %: 3.5 %
Minimum Detectable Relative Liftaccepts 1–200 %: 15 %

Results

Days Of Traffic Required: 25.6 days
Sample Needed Per Variant: 19,236
Total Sample Required: 76,945
Weeks Required: 3.66 weeks

Step by step

1Values used
2Sample size
3Days Of Traffic Required= 25.6 days
4Sample Needed Per Variant= 19,236
5Total Sample Required= 76,945
6Weeks Required= 3.66 weeks

Assumptions and limits

The result is derived only from the 4 inputs listed above. Anything not among them is not part of the calculation.
In particular, this calculator does not take inflation, tax, fees or charges or compounding frequency into account.
Figures are rounded for display to 2 decimal places, so totals built from several rounded parts can differ slightly from the unrounded arithmetic.
Figures are estimates based on the inputs given. Marketing performance, platform fees and conversion behaviour vary by audience, channel and season. Use this as a planning guide, not a forecast.
This is an estimate for general information and not professional financial, tax or legal advice. See the full disclaimer for the limits that apply to this kind of calculator.

How these calculators are built and tested is described on the Methodology page. If a figure looks wrong, please report it.

Figures are estimates based on the inputs given. Marketing performance, platform fees and conversion behaviour vary by audience, channel and season. Use this as a planning guide, not a forecast.

Frequently Asked Questions

Why does a small detectable effect cost so much traffic?

Because sample size scales with the inverse square of the effect — halving the effect quadruples the traffic needed.

Should the test run at least a week?

Yes, and ideally whole weeks, so weekday and weekend behaviour are both represented.