How many tests can your site actually run?

The hero answer rests on a few assumptions. Change any of them and the result above updates live. The defaults are deliberately conservative.

Assumptions (editable)

Reset to defaults

1. Sample size per test

Standard two-proportion z-test, using your assumptions above:

2. How many tests your traffic can power

The headline number is how many fully-powered tests your monthly traffic can feed:

Run one at a time on a single audience and a 7-day minimum caps you lower:

The headline is what your traffic could power if you run tests at the same time across separate pages or audiences. Back-to-back on one audience is the stricter floor. Running many overlapping tests on the same audience can go higher still, but risks tests interfering with each other.

3. The cost of testing below your ceiling

Each winner you don't run is a conversion lift you don't bank.

4. What a winning test is worth

We credit a realized lift per winner that's below the effect we test for, to stay conservative about regression to the mean and lifts that fade. This values each winner as a steady annual lift on your baseline, then stacks them.

Important: this is not guaranteed revenue. It's an indicator of the possible value of these tests if they run and win at conservative rates. Real lifts vary, and not every gain holds forever. Treat it as the size of the prize, not a promise.

Defaults are conservative, mid-market values. Edit them to match your own baseline, target lift, and value per conversion. The point is the shape of the answer, not a precise forecast.