Free tool
Survey sample size calculator
Find how many completed responses you need, how precise a sample you already have is, and how many people to invite. The maths runs in your browser.
Completed responses needed
385
At 95% confidence, ±5%, large population.
| Completes | Margin of error |
|---|---|
| 100 | ±9.8% |
| 200 | ±6.9% |
| 400 | ±4.9% |
| 1,000 | ±3.1% |
How this number was calculated
- n₀ = z² × p(1 − p) / e² = 1.96² × 0.50 × 0.50 / 0.050² = 384.2
- Large population: no finite-population correction.
- Round up: 385 completed responses.
Assumes a random sample from your population. Margin of error covers sampling error only, not non-response or question wording. Each subgroup you report on separately needs its own sample of roughly this size. Opt-in panels aren’t probability samples, so treat the margin as a guide.
Need 385 responses from US adults?
Get an instant quote for verified respondents and pay only for approved completes.
Get a quote →How to calculate a survey sample size
Choose a confidence level
95% means that if you repeated the survey many times, about 95 in 100 results would land within your margin of error of the true value.
Set the margin of error
Decide how far from the true answer you can afford to be. ±5% suits most decisions; use ±3% when options are close.
Add the population if it is small
If you are surveying a known group, such as 400 employees, enter it. The finite-population correction lowers the sample you need.
Read the required completes
That is the number of finished, usable responses, not invitations. Add a response rate to see how many invitations to send.
Repeat for each subgroup
If you will report regions or customer types separately, each one needs its own sample of roughly this size.
The formula
For a proportion, the sample needed from a large population is n₀ = z² × p(1 − p) / e², where z is the z-score for your confidence level (1.96 for 95%), p is the expected proportion as a decimal and e is the margin of error as a decimal.
For a known, smaller population N, apply the finite-population correction: n = n₀ / (1 + (n₀ − 1) / N). Results are rounded up, because you cannot collect part of a response.
Using p = 50% gives the largest sample, which is why it is the safe default when you do not know the answer in advance. At 95% and ±5% that is 385 completes for a large population.
Why a bigger population barely matters
Precision depends on how many people answer, not on the share of the population they represent. 385 responses give about ±5% whether the population is 100,000 or 300 million. The correction only bites when you sample a large share of a small group.
What the calculator cannot fix
Margin of error covers random sampling error only. It says nothing about who chose not to respond, leading question wording or a sample drawn from the wrong audience. A smaller, well-targeted sample is usually worth more than a large, skewed one.
Frequently asked questions
How many responses do I need for a survey?
For a large population, 385 completed responses give a ±5% margin of error at 95% confidence. About 1,068 give ±3%. For a small, known population, such as a 300-person company, the finite-population correction reduces the number.
Is 100 responses enough?
100 responses give roughly ±10% at 95% confidence. That is enough to spot large differences or directional trends, but not to separate options that are a few points apart.
What confidence level should I use?
95% is the convention for business and academic surveys. 90% is acceptable for low-stakes, directional work; 99% is for cases where a wrong call is expensive.
Does the calculator account for response rate?
The main result is completed responses. Enter an expected response rate to see how many invitations to send to your own list. For purchased respondents you pay per approved complete, so you order the completes directly.