All free tools

Zero Purchase Calculator

Estimate the chance of zero purchases at an assumed conversion rate, and the evidence a zero-purchase click sample provides.

Loading tool…

Put a zero-purchase result in context.

Zero purchases after a small number of clicks can occur even when the underlying conversion rate is healthy. Enter a conversion-rate assumption to calculate that probability. The tool also shows how much evidence a zero-purchase sample provides about the conversion rate, and how much you have spent relative to your target CPA.

Start with mature clicks

Use a click sample old enough for the normal purchase and reporting lag to pass. The sample in this calculator has zero observed purchases. Keep the conversion definition and click denominator consistent with your assumed rate. A store session rate and an ad click rate describe different samples.

Two calculations, two different questions

Probability of zero purchases = (1 − p)n, where p is the assumed purchase probability per click and n is the click count. Expected purchases are n × p. For a positive rate below 100%, the total clicks needed for at least a 95% chance of one or more purchases are the rounded-up value of log(0.05) ÷ log(1 − p).

The second calculation starts with the observed zero purchases. For n greater than zero, the exact one-sided 95% upper CVR bound = 1 − 0.051/n. This bound uses the sample count alone. Changing the assumed CVR changes the probability calculation, while leaving the confidence bound unchanged.

A worked example

In this illustrative example, you have 100 clicks, zero observed purchases, a $1 CPC and a 2% assumed click-to-purchase CVR. The expected purchase count is two. The probability of observing zero is 0.98100, or 13.26%. That outcome remains plausible under the 2% assumption.

The model reaches at least a 95% chance of one or more purchases at 149 total clicks. At the same CPC, that is $149 of total spend. The exact one-sided 95% upper CVR bound from the observed 100-click, zero-purchase sample is 2.9513%.

If your target CPA is $50, the $100 already spent equals two target CPAs. The forward scenario implies a $50 CPA from $1 ÷ 2%. Observed CPA remains unavailable because dividing spend by zero purchases has no finite result.

Use the result to choose a review

When a zero-purchase result is unusual under your assumption, review whether the sample is mature and the assumption still fits the traffic. Then inspect the offer, campaign, checkout and measurement separately. Compare the economics in the AOV and CVR planner or the cost cap planner. For delivery issues, use the Meta ads not spending guide.

Clicks without purchases questions

Does spending two target CPAs without a sale prove the ad is bad?

Compare the click count with an explicit conversion-rate assumption. The example above spends two target CPAs and still has a 13.26% probability of zero purchases. Spend exposure and probability answer different questions.

Does the 95% click count predict my next purchase?

It describes the total sample size that gives at least a 95% chance of one or more purchases under the fixed-rate model. It is measured from the start of a sample. It is not a countdown from your current clicks.

How should I read the upper conversion-rate bound?

It is the exact one-sided 95% confidence limit obtained by inverting the binomial zero-count probability. In repeated comparable sampling, the interval procedure covers the fixed true rate at least 95% of the time. A larger zero-purchase sample produces a lower upper bound.

What happens at a 0% conversion-rate assumption?

The probability of zero purchases is 100%. No finite number of clicks reaches a 95% chance of a purchase. The calculator still uses the observed click count to calculate its upper confidence bound when there is at least one click.

Can zero purchases tell me whether tracking is broken?

Check tracking through a separate event test and order reconciliation. A zero-count probability calculation uses the purchase counts you provide; it cannot identify why those counts occurred.

The binomial calculation assumes independent clicks with the same purchase probability and one binary purchase outcome per click. Repeat visitors, changing traffic, delayed purchases and attribution can violate those assumptions. References checked 16 November 2026: NIST binomial distribution and NIST exact binomial confidence intervals. The zero-count one-sided formula follows by solving (1 − p)n = 0.05.