Skip to content
TextArray1.117.0
100% local

Sample size calculator

Work out how many people to survey for a given confidence level and margin of error.

Input
Output
Related tools
A/B test significance calculatorNumbers
Percentage calculatorNumbers
Sum and number statisticsNumbers
Birth number validatorNumbers
Inline math evaluatorNumbers
Percent calculatorNumbers
Sample size calculator
Before running a survey or an experiment, you need to know how many responses actually make the result trustworthy. This calculator works out that number from three inputs: the confidence level you want, the margin of error you can live with, and the proportion you expect the answer to land near. Set the population size and it also applies the finite-population correction, which lowers the required sample once you are surveying a meaningful share of a small, known group—a company of 400 employees needs far fewer completed responses than an open web survey does.
By default the calculator assumes you are measuring a proportion, such as "percent who prefer option A," and 50% as the expected split, which is the safest assumption when you have no prior data because it produces the largest, most conservative sample size. Switch "Calculate for" to a mean instead, and it uses a standard deviation you supply—useful when the number you are estimating is an average (order value, time on task) rather than a percentage. The response rate option inflates the sample into a contact count: if only 40% of the people you reach out to typically respond, the tool tells you how many to invite to end up with the sample you actually need.
Paste one or more sample sizes into the input, one per line, and the tool runs the calculation in reverse: for each size, it reports the margin of error that sample size actually achieves at your chosen confidence level, with the same finite-population correction applied. Turn on "Show scenarios as a table" to compare several sizes side by side instead of reading them as separate lines—handy for a stakeholder deciding between a cheap, loose estimate and an expensive, tight one.
Everything runs locally in your browser: no survey plan, population figure or response data is ever uploaded anywhere. Copy the result, download it as a .txt file, or send it straight into another tool to keep working.
FAQ
What sample size should I use if I have no idea what the split will be?
Leave the expected proportion at 50%. It assumes the most variance and therefore produces the largest, safest sample size for any true split.
When does the population size matter?
It matters once your sample would be a meaningful fraction of the group you're surveying—an internal survey of 300 employees, say. For a large or open population, leave it at 0 and the correction has no effect.
What does the reverse calculation tell me?
Paste a sample size you already have (or are considering) and it reports the margin of error that size actually achieves, so you can see the trade-off before committing to a budget.
Should I calculate for a proportion or a mean?
Use a proportion for percentage-style answers (yes/no, prefers A/B). Use a mean, with a standard deviation, when you are estimating an average value like a rating or a duration.
Is my survey data uploaded anywhere?
No. Every calculation runs entirely in your browser—nothing about your survey or population is sent to a server.