Survey Response Rate Calculator

Response rate = completed responses ÷ invitations sent × 100.

Partial responses — people who started and abandoned — are shown separately. A large gap between the two figures means your survey is being opened and then abandoned, which is a length or question problem rather than a distribution problem.

The everyday formula is: response rate = completed responses ÷ people invited × 100. Send 1,200 invitations, get 288 completed surveys, and your response rate is 288 ÷ 1,200 = 24.0%. That is the headline figure the calculator shows with its default numbers.

The formula is simple; the arguments are about what goes on the top and bottom of the fraction. Does a half-finished survey count as a response? Does an email that bounced count as an invitation? Different answers can move the same survey by several points, which is why the survey research field agreed on standard definitions.

The American Association for Public Opinion Research (AAPOR) publishes the Standard Definitions, now in its 10th edition (2023). It sorts every invited case into completes (I), partials (P), eligible non-respondents (refusals and break-offs R, non-contacts NC, other O), and cases of unknown eligibility (U). Six response rates follow from how partials and unknowns are treated.

Before counting, decide what a complete and a partial are. AAPOR requires that decision up front and gives examples such as: more than 80% of applicable questions answered is a complete, 50–80% is a partial, and less is a break-off (which counts as a refusal, not a response).

For email and web surveys, two rules matter most. An undelivered invitation is not automatically ineligible: AAPOR puts it in unknown eligibility, because the person may still be in your target group. And if your list is accurate and current, people who simply never reply are eligible non-respondents, so they stay in the denominator.

Illustration. You email a customer survey to 1,200 addresses from a list you keep up to date. 40 bounce (unknown eligibility, U = 40). 12 reply that they are no longer customers (ineligible, removed entirely). Using an 80% / 50% rule, 288 are completes (I), 60 are partials (P) and 45 broke off early (R). The other 755 never replied, and because the list is current they count as eligible non-respondents.

Eligible known cases: 288 + 60 + 45 + 755 = 1,148. Add the 40 unknowns and you get 1,188.

RR1 = 288 ÷ 1,188 = 24.2%. RR2 = 348 ÷ 1,188 = 29.3%. With e = 0.5 (an assumption you would need to justify, for example from the share of bounced addresses you later confirm), the denominator is 1,148 + 20 = 1,168, so RR3 = 288 ÷ 1,168 = 24.7% and RR4 = 348 ÷ 1,168 = 29.8%. RR5 = 288 ÷ 1,148 = 25.1% and RR6 = 348 ÷ 1,148 = 30.3%.

Same fieldwork, rates from 24.2% to 30.3%. If you publish one, name the definition.

The calculator divides by the number you type in the invitations box, so you choose the denominator. Enter every invitation (1,200) and you get the simple rate: 24.0% completed, 29.0% including partials. Enter only eligible, known cases (1,148) and the two figures become RR5 and RR6: 25.1% and 30.3%. Enter 1,188 for RR1 and RR2.

Put only real partials in the partial box. If you also type break-offs there, the including-partials figure stops matching RR2, RR4 or RR6. In this example, entering all 105 starters who didn't finish would give (288 + 105) ÷ 1,200 = 32.8%, which no AAPOR rate endorses.

Response rate is measured against everyone you invited. Completion rate is measured against people who started: completes ÷ starts. In the example, 393 people started (288 + 60 + 45) and 288 finished, a completion rate of 73.3%, while the RR1 response rate was 24.2%.

The two point to different problems. A low response rate with a high completion rate means people aren't opening or clicking: sender, subject line, timing, list quality. A reasonable response rate with a low completion rate means the survey itself is losing people: length, a hard question, a broken page on mobile. The calculator's gap between completed and including-partials is a first signal of the second problem.

Precision depends on the number of responses, not the rate. 400 completed responses give roughly the same margin of error whether they came from 800 invitations or 8,000. Use the sample size calculator or margin of error calculator for that question.

What the rate indicates is risk of nonresponse bias: the chance that people who answered differ from those who didn't. That risk is real but loosely tied to the rate itself. Groves and Peytcheva's 2008 meta-analysis of 59 studies found response rates explained little of the variation in nonresponse bias between surveys. A 60% rate can be biased, and a 15% rate can be fine; compare respondents with what you know about the whole list (plan, region, tenure) to check.

It depends on the relationship and the channel: an internal staff survey and a cold external list are not comparable, and vendor reports often divide by views rather than invitations. Our survey response rate benchmarks article collects the published figures with their sources and denominators, so we don't repeat them here. The most useful comparison is your own previous send, calculated the same way.

You need three counts: invitations (from your email tool, minus confirmed ineligibles), completes and partials. In Zunoform, partial responses are saved as respondents move through the form (on by default), so you can count people who stopped without asking them. Each form's Summary tab also shows views, starts and completed submissions, which gives you the completion-rate side of the picture.

Resume works only in the same browser, so a partial is where that person stopped on that device. If you need one row per invited person, add a hidden field such as rid to each invitation link so you can match responses back to your list. To plan how many invitations to send in the first place, try the survey invitation planner.