Peak Atlas

Free tool · calculates in your browser only

Peak Forms sample size calculator

Choose a confidence level and margin of error, optionally add the population size, expected share and response rate: the calculator shows how many complete responses and invitations your survey needs.

Four brass instruments with pointers in front of burgundy paper strips – a metaphor for metrics and routines of manageable sales.

Your requirements

How certain you want to be that the true value lies within the margin of error. 95 % is common in practice.

How far a result may deviate from the true value at most, e.g. 5 for ±5 percentage points.

How many people could take part in total, e.g. all your customers. Leave empty if the number is unknown or very large.

The share you expect for your most important yes/no question. Empty means 50 %: this gives the largest and therefore most cautious sample.

Share of invited people who complete the survey. Only fill this in if you have your own figure from experience.

Inputs are processed only in your browser. Nothing is stored or transmitted. Either comma or period works as the decimal separator.

n₀ = z² · p · (1 − p) / e² · with a known population size N: n = n₀ / (1 + (n₀ − 1) / N)

Complete responses needed

–

Rounded up. This is how many completed responses you need for the chosen margin of error and confidence level.

z-value used

1,960

Two-sided quantile of the standard normal distribution for the confidence level.

Assumption: The formula applies to proportions (e.g. “How many are satisfied?”) and a random selection from the population. No sample size makes up for bias, for example when only very satisfied people respond.

Method and sources: Normal approximation with finite population correction following W. G. Cochran, Sampling Techniques, 3rd edition, Wiley 1977, chapter 4. See also NIST/SEMATECH e-Handbook of Statistical Methods.

Next stepCreate a survey with Peak Forms

Limits

What the calculator doesn't answer.

  • The formula applies to proportions from a random sample. Means, group comparisons or A/B tests need different calculations.
  • A large enough sample does not protect against bias: if certain groups respond less often, the result stays skewed.
  • If you want to analyse subgroups (e.g. by region), you need the calculated number per subgroup, not in total.

Next

Goes well with this.

Conversation

You have the numbers but not the decision? An initial consultation puts them in context.

Analyze my shop

Start with Clarity. The rest follows.

Sign in with Google, build your atlas in about two minutes and see your company in one place.

Free plan · Sign in with Google · set up in about two minutes