Enter population, confidence level, margin of error, and (optionally) your expected proportion.
Input values
Result
Already have responses? Find out how accurate your results are.
Input values
Result
The number the calculator gives you is completions needed — not invitations sent. If your expected response rate is 20% (typical for email surveys), multiply the result by 5 to get your invitation count. We've seen teams launch surveys after hitting the sample size target, not realising 30% of responses were incomplete. Build in a 15–20% buffer and filter for complete responses before calculating.
What the inputs mean
Population size is simply the total number of people you could survey. That might be your customer list, your employee headcount, or the attendees at an event. If you genuinely do not know — say you are running a public poll — leave it blank and the calculator uses the conservative infinite-population formula.
Confidence level and margin of error work as a pair. Think of them as a deal you make with uncertainty. 95% confidence with a ±5% margin of error means: if you ran this survey 100 times, 95 of those runs would produce a result within 5 percentage points of the true answer. That is the clearest explanation that exists, and the one worth remembering. If you do not know what percentage of people will answer “yes”, leave the expected proportion at 50%. It gives you the largest, most conservative sample size, which is always the safest place to start.
How many responses do I actually need?
The answer depends entirely on who you are surveying.
For internal or employee surveys, your population is typically 50 to 500 people. At this scale, the finite population correction matters a lot. Surveying 100 people from a company of 200 is very different from 100 out of 10,000. The calculator output changes significantly once the population drops below 1,000, so entering your exact employee count is worth the extra 10 seconds.
Customer feedback surveys sit in the middle — populations of 1,000 to 100,000. The standard 95% confidence with ±5% margin benchmark gives roughly 370 responses regardless of whether your customer base is 10,000 or 100,000. The curve flattens out once the population gets large, which surprises most people. Three hundred and seventy well-targeted responses tells you as much about 100,000 customers as it does about 10,000.
For market research or public polls, the population is large or unknown. Use the 50% proportion, 95% confidence, ±3% margin for serious research. That gives you roughly 1,067 responses. ±5% is fine for directional decisions. ±3% is the threshold where you can confidently say one option beat another by a meaningful margin. The calculator handles all three scenarios — just enter what you know and leave the rest blank.
Invitations vs responses
The sample size output is completions needed, not people to contact. This is the most commonly misunderstood step. To hit 370 completions with a 20% response rate, you need to send 1,850 invitations. Email surveys typically see 15 to 30% response rates. In-app surveys tend to land between 10 and 20%. SMS sits at 25 to 45%. Internal employee surveys run 30 to 60%, depending on how senior leadership frames the ask.
Add 15 to 20% to your target to account for incomplete responses — people who start but do not finish. Filter for complete responses before you analyse. The numbers the calculator produces are based on clean, complete data. Incomplete responses do not count toward your sample size target.
What to do once you have your number
Check your actual list size against the population input you used. If your CRM says 8,247 active customers, use that exact number — rounding up to 10,000 costs you nothing but gives you a slightly more conservative answer. Plan your distribution channel based on the expected response rate, then calculate invitations needed by dividing your target completions by that rate. If you cannot reach the target, lower your precision — widen the margin of error — rather than report misleading results. An honest ±7% beats a dishonest ±5% every time.
Need help collecting responses? See our guide to improving survey response rates or compare survey tools by response rate features.
