Customer satisfaction questions to ask, by moment

Pick the moment first, then ask one 1 to 5 satisfaction question about it and one optional open question. Most surveys need one to three questions, so the list below is a menu to choose from, not a form to copy whole.

If you only need the support version, use the CSAT survey questions template. For the full method of running the survey, see how to run a customer satisfaction survey.

Start with the moment, not the list

A satisfaction question works when it names one thing that just happened. Survicate looked at 835 CSAT surveys from 269 of its own customers and found that 32.2% of the questions asked about the company overall instead of a specific interaction. ¹ The sample is not a benchmark, but the habit is easy to spot in your own surveys.

Before you add a question, ask what the team will do differently if the score is low. Survicate suggests that if nobody has an answer, the question should not be there. ¹

Keep it short. Qualaroo advises one scored question plus an optional open follow-up, and says a survey over four or five questions is too long. ³ Typeform links surveys that take seven minutes or more to higher abandonment. ²

Questions by moment

The tables mark each question as a rating (R, 1 to 5 with every point labeled) or open (O). Pick one row per moment as your starting point.

First week

Typeform suggests asking about onboarding within the first week, while it is still fresh. ²

After a purchase

After a support conversation

This moment already has owners on this site. The CSAT template linked at the top has the rating and its follow-ups, and open-ended survey question examples lists follow-ups by score. If you use Zendesk, send the survey when a ticket is solved.

For effort rather than satisfaction, use the CES template.

Mid-relationship, every few months

Qualaroo places NPS at quarterly or post-renewal loyalty checks. ³ The NPS, CSAT and CES comparison explains when each one fits.

When someone cancels or goes quiet

Typeform pairs "What almost stopped you from continuing?" with customers who stayed after a renewal. ² It also names re-engagement as a time to ask. Treat both as optional.

A worked example

Made-up example: a 30-person software company sends two short surveys. Every number below is invented to show the sums.

After onboarding (two questions): "How clear was getting started?" (1 to 5) and an optional "What was the hardest part?"

Forty people answer. Two give a 1, three a 2, five a 3, fourteen a 4 and sixteen a 5. The 4s and 5s are 30 of 40 answers, so CSAT is 75%.

The team reads the five low ratings first. Three of the five comments say they could not find where to invite teammates. That is one concrete change: add "Invite a teammate" to the setup checklist.

After support (two questions): "How satisfied were you with the help you received?" and an optional "What is the main reason for your rating?"

Twenty-five people answer: one 1, two 2s, three 3s, nine 4s and ten 5s. That is 19 of 25, so CSAT is 76%.

The two scores are similar, but they point at different fixes, because each one names a moment. A single "How satisfied are you with us?" question would have given one number and no clear change.

To check your own sums, use the CSAT calculator. To combine several ratings like these into one 0 to 100 number, see the customer satisfaction index guide.

Questions to leave out

• Details you already hold. Plan, country, company size and signup date are often already in your own data and can travel with the answer, so do not ask for them.
• Two questions in one. "How satisfied were you with our speed and quality?" cannot be answered with a single number.
• Leading wording. "How great was your experience?" pushes the answer. Offer a neutral scale with every point labeled.
• "Overall, how satisfied are you with us?" after a single event. It mixes everything you have ever done into one number. Name the event.
• "Anything else?" on every survey. Keep it for the last question of the cancellation survey. Elsewhere, one open question is enough.

How often to ask the same person

Count the moments, not the questions. Take a customer who onboards, files a ticket and reaches a quarterly check-in. Each moment gets one short survey, so over the months they answer three surveys of one to three questions each, and none of them asks the same thing twice.

The rules for gaps between surveys are in how to ask for feedback without annoying people. In Pulseri, the minimum gap between relationship surveys to one person is 90 days by default, and you can send an event survey, such as one after a Zendesk ticket is solved, with a set delay. The survey builder covers the setup.

Low ratings are worth reading the same day, and the open answers are worth a weekly review, which voice of the customer for small teams lays out. A Slack alert for low scores sends them to the team as they arrive, and Pulseri can alert Slack or Microsoft Teams. For the definition behind the score, see what CSAT means.