The survey sample size calculator.
Two answers in one place: how many people a study needs, and what margin of error a sample already gives you. Every number updates live, and nothing you type leaves your browser.
How many people do I need to survey?
How sure you want to be. 95% is the usual survey standard.
Everyone you want to speak about. Blank is fine: above about 100,000 it makes almost no difference.
Survey about
385 people
to land within ±5% at 95% confidence, for a large or unknown population.
Pick a margin of error target
| Margin of error wanted | Responses you need |
|---|---|
| ±1% | 9,604 |
| ±2% | 2,401 |
| ±3% | 1,068 |
| ±5% | 385 |
| ±7% | 196 |
| ±10% | 97 |
Everything runs in your browser. Nothing you type leaves your device.
Rule of thumb: 385 is the number to remember.
At 95% confidence, 385 responses from a large population lands you within about ±5%. It’s not folklore: 1.96 squared, times a quarter, divided by 0.05 squared, rounds up to 385. Halving the margin roughly quadruples the sample, so decide how much wiggle room you actually need before you promise the board a census.
What does a confidence level actually mean?
Horizontal position is distance from the true value, in standard deviations (z-scores).
The middle 95% of the curve is the region you trust. The two pink tails share the remaining 5.0% between them: the chance a properly drawn sample still lands far enough out to mislead you. Push the slider toward 99% and the trusted middle widens while the risky tails shrink. Certainty costs width.
Your estimate can sit above the truth or below it, and both count. That’s why the pink tails sit on both sides of the curve: the margin of error guards against misses in either direction. Drag the slider above and watch it happen live. More confidence means a wider trusted middle and slimmer risky tails.
The words, in plain English.
- Population
- Everyone the study is meant to describe: all your users, all shoppers in a city, all adults in a country. Big populations are surprisingly cheap to speak about, which is why the sample barely changes as they grow.
- Sample
- The slice of the population you actually survey. The whole gamble of research is that a well-picked slice behaves like the whole. Random slices behave. Hand-picked ones flatter.
- Confidence level
- How often your method would capture the truth if you reran the study loads of times. Pick 95% and, in theory, 19 intervals out of 20 will contain the real value. It’s a promise about the method, not about any single result.
- Margin of error
- The wiggle room around your result. If 58% of your sample likes the new design and the margin of error is 5%, the honest headline sits somewhere between 53% and 63%.
- Confidence interval
- The margin of error written as a range: your result, minus and plus the wiggle room. A finding of 62% with a 4% margin of error has an interval running from 58% to 66%.
What else affects how accurate my survey is?
The calculator assumes a random sample, the honest ideal where everyone in the population has a fair shot at being asked. Real studies fall short of that, and the fix is rarely a bigger respondent count. Careful survey and questionnaire design beats padding the sample every time.
Curious how we would run yours? See the nine-step approach, read the quick guide to UX research, or start with how market research and UX research differ.
Questions the calculator gets asked.
For a general-population survey at 95% confidence and a 5% margin of error, about 385 responses. Tighter margins cost sharply: 3% needs roughly 1,068, and 1% needs over 9,600. Enter your actual population in the calculator and it adjusts from there.
At 95% confidence, yes, for a large population. It comes straight from the formula, not folklore: 1.96 squared, times a quarter, divided by 0.05 squared, rounds up to 385. Smaller populations need slightly fewer responses, and the calculator applies that correction for you.
Barely. Whether your population is a hundred thousand or a hundred million, 5% at 95% confidence still costs about 385 responses. The sample you need depends on the precision you want, not on how many people are out there.
5% is the working standard for most surveys. Choose 3% when the findings feed big spending decisions, and accept 7% to 10% for quick pulse checks. Below 3%, sample sizes get expensive quickly for very little extra certainty.
Margin of error shrinks with the square root of your sample, so gains taper off. Going from 100 to 400 responses is a big win. Going from 4,000 to 4,300 barely moves the needle. The chart shows the flattening clearly.
No. All the maths runs in your browser on your device. Nothing you enter is transmitted, stored, or logged, which keeps the tool comfortable to use under POPIA.
Calculator reviewed October 2026.
Let's talk
Ready to stop guessing?
Tell us what you're trying to figure out. We'll design the study that answers it: UX, market, or both.