Survey response rate

A score is only as trustworthy as the group that produced it. If 2% of invited customers answer, the results describe that 2% — often the very happy and the very unhappy — not the customer base. Response rate is therefore less a performance metric than a confidence measure: it tells you how much weight the other numbers can carry.

It is also an early warning signal. A falling response rate usually appears before anything else goes wrong: customers tiring of surveys, a broken link, an offline kiosk, an SMS sender being filtered.

Response rate = (completed responses ÷ invitations delivered) × 100

Example: an email survey goes to 4,000 customers; 200 bounce, so 3,800 are delivered. 950 people open it, 400 start it and 340 finish. Response rate = 340 ÷ 3,800 × 100 = 8.9%. Completion rate (finished ÷ started) = 340 ÷ 400 = 85%.

Metric	Formula	What it diagnoses
Delivery rate	Delivered ÷ sent	Data quality, sender reputation
Open or click rate	Opened ÷ delivered	Subject line, sender, timing
Response rate	Completed ÷ delivered	Overall effectiveness
Completion rate	Completed ÷ started	Survey length and design

For kiosks and QR codes there is no invitation list, so teams track responses per visitor (responses ÷ footfall or transactions) or simply responses per day per location, and watch the trend.

• Channel and moment. A kiosk at the exit or a QR code at the table catches customers while the experience is fresh; an email two days later competes with everything else in the inbox.
• Length. Every extra question loses some respondents. Transactional surveys should take under a minute.
• Relevance. “How was your visit to our Riverside branch today?” outperforms “Please take our customer survey”.
• Language. Customers answer more readily in their own language.
• Trust and visible action. Customers who see feedback lead to change keep answering.
• Frequency. Asking the same customer too often lowers response — see survey fatigue.

1. Cut the survey to the questions you will act on.
2. Ask in the moment — kiosk, QR or an SMS within an hour.
3. Use one tap to start — put the first rating question directly on the screen or in the message.
4. Offer the survey in the customer’s language with a visible language switch.
5. Personalise the invitation with the branch, date or service.
6. Close the loop visibly — “You told us, we changed it” messages keep customers engaged.
7. Test variants. Compare two versions of the survey or invitation and keep the better one.

Qmeter supports these levers: kiosk, QR, web, SMS and email channels; surveys in more than 100 languages with AI-assisted translation; A/B tests of two to five survey variants with a winner chosen after a set number of responses; and campaign reports with opens and clicks. See multilingual surveys and survey campaigns.

A high response rate is reassuring, but what ultimately decides whether a score is reliable is the number of responses behind it. For a percentage such as top-2-box CSAT, a common rule of thumb for the margin of error at 95% confidence is roughly 1 ÷ √n, where n is the number of responses:

Responses (n)	Approximate margin of error
100	± 10 percentage points
400	± 5 percentage points
1,000	± 3 percentage points

So a branch with 100 responses a month that moves from 78% to 72% may not have changed at all, while the same move over 1,000 responses is a real signal. NPS, which subtracts two percentages, has a wider margin than a single percentage. When volumes are low, report longer periods or combine similar locations rather than reacting to monthly noise.

• Dividing by invitations sent rather than delivered, which hides list-quality problems.
• Comparing response rates across channels as if they were the same thing.
• Chasing volume with incentives that attract answers from people who did not use the service.
• Ignoring low-response branches, where a handful of answers can swing the score wildly.