By the Wavefield Research team · Published Sep 21, 2026
A good survey response rate is 10–30% for external email surveys, 25–40% for customers with a real relationship, and 50%+ for employees. But the number everyone asks about is the wrong worry: what damages a survey is not a low rate — it is who the missing people are.
This page gives the benchmarks straight, defines the calculation the way methodologists do (most published rates are computed loosely), and then makes the argument the benchmark listicles skip — the one that changes what you should actually do about a low rate.
What counts as a good response rate for a survey depends mostly on who is being asked — audience relationship drives response more than any design choice. The ranges below assume email invitations and a survey under ten minutes.
10–15%
Cold-ish consumer lists sit at the bottom of the range; a clean, recent list with a real sender name reaches the top of it.
15–30%
Professional audiences respond better when the topic touches their work — and worse than consumers when it doesn't.
25–40%
A genuine relationship is the strongest response driver there is. Below 20% with your own customers usually means list decay or survey fatigue, not a survey problem.
50–70%+
Employee surveys carry an implicit expectation to participate. The number to watch is not the rate but who the non-responders are — the disengaged leave the survey unanswered first.
Not comparable
Panel studies fill quotas rather than chase rates: the vendor invites until the cells fill. Judge panel work by quality checks and incidence, not response rate.
SMS invitations run meaningfully higher than email for consented lists — texts get opened — with the consent mechanics covered in SMS surveys. And every range above measures invitations you control; sample you purchase fills quotas instead, which is why the response-rate question mostly disappears the moment a panel enters the design.
The formula everyone quotes — completes divided by invitations, times 100 — hides three decisions that change the number. Bounces leave the denominator: an invitation that never arrived is not a refusal, so 1,000 sent minus 80 bounced is a denominator of 920. Partials stay out of the numerator: a respondent who quit at question 12 is an abandonment, tracked separately, not half a complete. Screen-outs are their own category: they responded and didn't qualify — counting them as non-response understates engagement, counting them as completes inflates it. AAPOR's Standard Definitions formalize exactly these rules into numbered response-rate formulas — the existence of six of them is itself the lesson that an unqualified “response rate” is an ambiguous claim.
The planning direction matters more than the reporting one: completes needed divided by expected response rate equals invitations required. A study needing 385 completes at a 25% rate means inviting roughly 1,540 people — the arithmetic worked through in the sample size calculator. On Wavefield the field monitor does this bookkeeping live: invited, started, completed, abandoned, and screened out are separate counts, with abandonment points showing exactly which question loses people.
The advice industry treats response rate as a proxy for validity — higher rate, better survey. The best evidence says the relationship is far weaker than intuition insists: Pew Research Center's work on low response rates found telephone surveys answering in the single digits still produced accurate estimates across most measures. The mechanism: nonresponse bias arises when the propensity to respond correlates with the thing being measured — not from the rate itself. A 6% response from a sample whose responders and non-responders hold similar views measures fine. A 70% employee survey whose silent 30% are precisely the disengaged employees is badly biased at more than ten times the rate.
The practical program follows from the mechanism. Compare your respondents' demographics against the population you invited — deviations are visible and correctable by weighting. Ask whether the survey's topic itself filters who answers (a satisfaction survey oversamples the delighted and the furious — the known U-shape of feedback response). And report the rate with the result rather than hiding it: a stated 12% with demographic checks is more credible than an unstated anything. Chasing the rate for its own sake buys less validity per dollar than checking and correcting who actually answered.
Ranked by reliability, not novelty. One reminder to non-responders is the most dependable lever in fieldwork — it reaches the people who meant to respond and forgot, which is most non-responders on a warm list. Send it two to four days after the invitation; on Wavefield the reminder targets pending contacts automatically, so completed respondents are never re-nagged. Length honesty comes second: response rate is decided at the invitation, completion rate inside the survey, and a stated honest length (“7 minutes”) protects both — while the field monitor's abandonment points show which question is bleeding people when completion is the problem. Sender identity third: a recognizable name and a subject line naming the topic beat every clever alternative.
Incentives work but change who responds — a drawing or credit pulls in respondents with weaker topic engagement, which is sometimes exactly what you want (less self-selection) and sometimes noise. And the structural lever the tips lists never mention: survey fewer people, less often, with surveys that visibly led to something. Response rate is a reputation your surveys earn over time with the same audience — the fastest way to a bad rate is a history of long surveys that changed nothing, a discipline covered in customer feedback surveys.
Working ranges: 10–15% for external consumer email, 15–30% for B2B, 25–40% for your own customers, 50%+ for employees. Anything above 30% on an external list is genuinely strong. But the rate is a fieldwork health metric, not a validity certificate — a survey's accuracy depends on whether the people who didn't respond differ from those who did on what you're measuring, which is why representativeness checks matter more than hitting a threshold.
Published averages for online surveys cluster between 10% and 30%, with the typical survey response rate for external email invitations landing near 20%. Averages hide the drivers: audience relationship (customers respond several times more often than cold lists), survey length (completion drops measurably with every added minute), and whether a reminder went out. Compare your rate to your own past waves on the same audience — that comparison is diagnostic; the global average is trivia.
Completed surveys divided by delivered invitations, times 100. The details are where numbers get gamed: bounced emails come out of the denominator (they never received an invitation), partial responses do not count in the numerator, and screen-outs are their own category — they responded but didn't qualify, so report them separately rather than as non-response. AAPOR's Standard Definitions exist precisely because 'response rate' without these rules can mean six different numbers.
No — and this is the most persistent myth in survey research. Pew Research Center's studies found telephone polls with single-digit response rates still produced accurate estimates on most measures, because bias depends on whether responding correlates with the thing being measured, not on the rate itself. A 6% survey of a well-mixed sample can beat a 70% survey whose missing 30% all share a viewpoint. Check who is missing, weight what you can, and report the rate honestly.
Related: sample size calculator · survey weighting · customer feedback surveys · data collection methods
Invited, started, completed, abandoned, screened out — tracked live, with one-click reminders to exactly the people who haven't answered. From $99 per study.
Start a study