Glossary · Measurement
Response rate
Response rate is the number of usable responses divided by the number of people invited to take part, expressed as a percentage. It measures how much of your intended sample you actually reached. It is a different number from completion rate, which measures only how many of the people who started the survey got to the end, and a survey with no known invitation list has no meaningful response rate at all.
The formula, and the hard part
Response rate = usable responses ÷ people invited × 100.
The numerator is easy. The denominator is where most reported response rates fall apart, because "invited" has to mean something specific. If you emailed 4,000 addresses and 300 bounced, your denominator is 3,700, not 4,000. If you posted a link on social media, you do not have a denominator at all, and you cannot report a response rate — only a response count.
Be explicit about which you are quoting. A survey with 1,200 responses from an open link is a perfectly respectable piece of work; describing it as having a response rate is not.
Response rate versus completion rate
These get used interchangeably and they measure different failures.
| Response rate | Completion rate | |
|---|---|---|
| Denominator | People invited | People who started |
| Measures | Reach and motivation to begin | The survey itself |
| Fix it by | Better invitation, timing, reminders, incentive | Shorter survey, better logic, fewer free-text questions |
A low response rate with a high completion rate means your survey is fine and nobody opened it. A high response rate with a low completion rate means your invitation worked and your survey is too long. See completion rate for the second half of that picture.
A worked example
A trust emails 5,000 staff. 180 addresses bounce. 1,100 people open the survey and 860 submit a response, of which 12 are blank test entries you clear out.
- Denominator: 5,000 − 180 = 4,820 delivered invitations.
- Usable responses: 860 − 12 = 848.
- Response rate: 848 ÷ 4,820 = 17.6%.
- Completion rate: 848 ÷ 1,100 = 77%.
Two very different improvement projects sit behind those two numbers.
What actually moves it
- Who is asking. An invitation from a named person people recognise beats one from a no-reply address, reliably.
- An honest length estimate. "About four minutes" that turns out to be twelve costs you more than saying twelve would have.
- Reminders. One reminder to non-responders is usually worth more than any amount of rewriting the original email.
- Telling people what happened last time. "Last year 900 of you told us X, and we changed Y" is the single most under-used lever in public and staff consultation.
- Removing barriers. No login, works on a phone, and a resume link for anything long.
Why it matters more than the number suggests
Response rate is not just a productivity measure; it bounds how much you can trust the results. A low rate is only a problem if the people who did not respond differ systematically from the people who did — but you usually cannot know that, and the safe assumption is that they do. That is sampling bias, and it is the reason a 17% response rate with a broad, representative spread beats a 60% rate from one department.
In NumoForms
Results show response counts and completion per survey. The invitation denominator lives with you, because the product does not run your mailing list — so record how many invitations you sent and when, at the time you send them, or the number is unrecoverable afterwards. If you distribute through several channels, a hidden field in each link lets you split responses by source and work out which channel is carrying the survey.
Related terms
Completion rate
Completion rate is the share of people who started a survey and reached the end. What counts as a start, where drop-off clusters, and how to bring it back up.
Sampling bias
Sampling bias is systematic error from who ends up in your sample. Self-selection, coverage error, non-response, and why a bigger sample fixes none of it.
Conditional logic
Conditional logic shows or hides a survey question based on an earlier answer. How any/all matching works, and what it does to exports and scoring.
Screener question
A screener question decides whether a respondent belongs in the study before the survey proper begins. Where to put it, how to word it, and what it cannot fix.
Back to the survey glossary, or read the product overview to see how these ideas map onto the builder.
Give people fewer reasons to stop.
Responses need no login, work on a phone, and anything long gets a resume link, so a half-finished answer is not a lost one. Results show counts and completion, and drop-off analytics tell you the question people leave at.
- Save and continue, with a resume link
- Drop-off analytics, question by question
- Hidden fields to record which channel each response came through
