A sample size calculator tells you how many people you need to survey for your results to be reliable. Choose how confident you want to be, and decide how much error you can accept. This sample size calculator then gives you the number of completed responses to aim for. If your group is small, such as the staff of one company, it adjusts the figure downward. Add a likely response rate as well, and it works out how many people to invite. It's designed for surveys that estimate a proportion or a percentage.
How the sample size calculator works
The tool starts with the standard formula for estimating a proportion, where Z is the score for your confidence level, p is the expected proportion and e is the margin of error:
n₀ = Z² × p × (1 − p) ÷ e²
If you enter a population size N, it then applies a finite population adjustment:
n = n₀ ÷ (1 + (n₀ − 1) ÷ N)
Results are always rounded up to the next whole person.
| Confidence level | Z-score | Sample at ±5% | Sample at ±3% | Sample at ±1% |
|---|---|---|---|---|
| 90% | 1.645 | 271 | 752 | 6,766 |
| 95% | 1.96 | 385 | 1,068 | 9,604 |
| 99% | 2.576 | 664 | 1,844 | 16,590 |
How to use the calculator
- Choose a confidence level. 95% is a common choice.
- Enter your margin of error, such as 5%.
- Enter a proportion if you have a good estimate.
- Enter your population if the group is limited.
- Add a response rate to see how many people to contact.
- Press Calculate.
A worked example
Suppose you want to survey the 10,000 customers of a business, with 95% confidence and a 5% margin of error. Using a proportion of 50%, the first step gives 1.96² × 0.5 × 0.5 ÷ 0.05², which is 384.16. The population adjustment then gives 384.16 ÷ (1 + 383.16 ÷ 10,000), or about 370. You need 370 responses. If you expect that 20% of the people will reply, you should invite 1,850 customers.
What do these terms mean?
Sample size determination is the act of choosing the number of observations to include in a statistical sample. A larger sample gives you more precise results, but it also costs more in time and money, and in practice the size is often determined by those limits. The margin of error is the plus-or-minus range around your result. If 60% of respondents say yes, and your margin is 5%, the true figure is likely to be between 55% and 65%. The confidence level describes how often an interval made in this way would contain the true value.
Why 50% is the safe default
The term p(1 − p) is largest when p is 0.5, where its value is 0.25. That means that 50% gives you the biggest sample size, so it's the conservative choice when you don't know what the answer will be. If you have a reasonable estimate from earlier research, you can use it instead. For example, using 20% at 95% confidence and a 5% margin reduces the sample from 385 to 246 responses.
Does population size matter?
Less than most people expect. The margin of error is essentially the same regardless of whether the population of interest is a school, a city, a state or a country, as long as the sampling fraction is small. A finite population correction starts to make a real difference when you're sampling more than about 5% of the group. That's why a survey of a whole country can be just as precise with roughly the same number of people as a survey of one large city.
Why margins are so costly
Precision becomes expensive quickly, because the margin of error is squared in the formula. Cutting your margin in half makes the sample about four times as large. At 95% confidence, a 5% margin needs 385 responses, and a 1% margin needs 9,604. Think carefully about how much precision your decision really needs.
When the numbers aren't enough
The formula assumes that your sample is chosen at random. If only your happiest customers reply, a very large sample will still give you a misleading answer. Try to reach people in a way that doesn't favor any particular group, keep your questions neutral, and follow up with those who don't respond.
Choosing a confidence level
A higher confidence level gives you more assurance, but it also takes more responses. Moving from 90% to 95% at a 5% margin raises the sample from 271 to 385, and moving up to 99% raises it to 664. For an informal poll of customers, 90% may well be enough. For research that will shape an important decision, 95% or 99% is easier to defend. Whichever level you choose, report it along with your margin of error, so that readers can judge how much weight your findings deserve. A result quoted without those two numbers is hard to interpret. Decide on both before you start collecting answers, because choosing them afterward invites the temptation to pick whatever makes the result look strongest. It also helps to plan for incomplete answers, since some people will start your survey and leave without finishing, and those partial responses may not be usable.
Common mistakes
- Counting invitations as responses. Only completed replies count.
- Using a formula meant for proportions on averages. Comparing means requires a different calculation.
- Ignoring subgroups. Each group that you plan to report on needs enough responses.
Related statistics tools
Analyze your results with the standard deviation calculator, work out odds with the probability calculator, and find the share of an answer with the percentage calculator.
What next
Use this often? Add Sample Size to your Home screen.
Frequently asked questions
How does the sample size calculator work?
It applies the standard formula for a proportion, using your confidence level and margin of error, and it adjusts for small populations.
How many responses do I need for a survey?
At 95% confidence and a 5% margin, about 385 for a very large population.
Why use 50% for the proportion?
It gives the largest and most cautious sample size when the true answer isn't known.
Does a bigger population need a bigger sample?
Only slightly, unless you're sampling a sizable share of the group.
What if people don't respond?
Enter a response rate, and the tool shows how many people you should invite.
Was this tool helpful?
What went wrong?
We only receive the reason you pick, never what you typed into the tool.
Thanks for your feedback.