A customer satisfaction survey is a short question sent right after a specific moment, such as a solved support ticket, that asks how satisfied the customer was. Four choices decide whether the result is useful: the moment, the question, the length and the delay. This post sets a default for each, then covers how many invitations to send and what to do with the answers.
The CSAT glossary entry has the short definition, and the guide to the customer satisfaction score has the formula. For ready-made wording, use the CSAT survey question template.
Pick the moment before the question
Ask about one moment, not about the company in general. A person can rate a support chat they had yesterday. They cannot rate "your experience" without averaging a year of memories. For a rating of the whole relationship, NPS is the better fit.
Pick the moments where a bad experience costs you something and where you can still fix it. Four common ones:
| Moment | When to send | Starting delay (our suggestion) | Who follows up |
|---|---|---|---|
| Support ticket solved | When the ticket status changes to solved | 1 hour, or a day if tickets are often reopened | The support lead |
| Onboarding finished | When the customer completes setup | Same day | Customer success |
| Purchase delivered or first use | After the customer has had time to use it | A few days after delivery | The product owner |
| Renewal or plan change | After the change takes effect | A few days | The account owner |
The delays are suggestions to start from, not statistics. Change them when your own response rate or comments tell you people answered too early or too late.
Ask one question on a five-point scale
Use one rating question tied to the moment: "How satisfied were you with your recent experience?" Answers run from 1 to 5, and 4 and 5 count as satisfied. The score guide above explains how the scale and the cut-off change the number. For the survey, pick one scale and keep it. Pipefy, which sells workflow software, describes labeled points from extremely dissatisfied to extremely satisfied. 2
Show a comment box after the rating, not before it. Pipefy suggests an open-text follow-up when someone gives a 1 or a 2. 2 People who are happy finish in one tap, and people who are not can tell you why. For wording of that comment question, see open-ended survey questions with examples.
Keep it short
Longer surveys tend to be finished less often. Survicate, a survey vendor, reports an average completion rate of 83.34 percent for surveys with one to three questions, the highest of the lengths it compares. 1 It does not say how it measured this, so read the figure as a direction, not a promise.
One rating question and one comment box is a short survey. If you also want to know about price, features or the competition, run a separate survey at a moment that fits each topic. A question about a competitor, for example, does not belong in a survey sent an hour after a support ticket.
Send it at the right time
Send soon enough that the customer remembers the moment, and late enough that the moment has ended. Survicate suggests asking soon after a purchase, and waiting two weeks to a month before questions that compare you with competitors. 1 For support, ask once the ticket is closed, not during the conversation.
When the moment ends with a status change, add a short delay so a reopened ticket does not turn a rating into noise. The Zendesk CSAT setup guide covers that case, including how to pick the delay and how often one customer can be asked.
Work out how many invitations to send
The number of answers you need sets the number of invitations, and the response rate links the two. The score guide shows how few answers make a percentage unreliable.
Made-up example: you want about 100 answers per month for one moment. Divide 100 by your response rate to get the invitations to send.
| Response rate | Invitations for 100 answers |
|---|---|
| 10 percent | 1,000 |
| 20 percent | 500 |
| 30 percent | 334 |
Measure the rate for each moment during the first month. It will differ between a support ticket and a renewal, so a single figure for all of them can leave one moment with too few answers.
Act on the answers
A survey that nobody reads teaches customers to stop answering. Decide who looks at the results before you send the first one.
For a low score, a short routine works:
- Read the comment the same day, while the case is fresh.
- Give one person the follow-up, so it does not fall between teams.
- Reply to the customer, even if the fix will take weeks.
- Note the cause. After a month, count the causes and fix the most common one.
Compare the overall score only with your own earlier scores for the same moment and the same scale. A post-purchase score and a support score measure different things.
To check a score quickly, use the CSAT calculator.
Run it in Pulseri
Pulseri's CSAT survey asks "How satisfied were you with your recent experience?" on a 1 to 5 scale, and the score is the share of 4s and 5s. It is part of customer feedback surveys, which also cover NPS and CES. A triggered campaign can send when a Zendesk ticket is solved, with a delay you set in seconds. The Zendesk trigger is not on every plan, so check the pricing page first.
Sources
- Customer satisfaction survey questions (Survicate) (retrieved 4 Nov 2026) Back to text: 1, 2
- What is a customer satisfaction (CSAT) survey? (Pipefy) (retrieved 4 Nov 2026) Back to text: 1, 2



