The Zendesk Satisfaction Survey
The Zendesk satisfaction survey is one question, sent automatically after a ticket is solved. Setup takes ten minutes, and the timing moves your score more than your agents do.
What the Zendesk satisfaction survey actually is
Turn satisfaction ratings on in the admin settings and Zendesk starts emailing requesters after their ticket is solved. The classic form is a single question, good or bad, with an optional comment box.
The answer writes back to the ticket as a satisfaction value, which is the part that matters. Because it lives on the ticket, you can slice it later by agent, group, channel, form, brand and tag without doing anything special.
Some plans expose an expanded rating scale and extra survey options beyond the two-button version. That area has changed more than once, so check the current Zendesk docs for what your tier gives you before you design a programme around a specific format.
One thing to know up front: the survey goes to the requester and nobody else. CCs and followers aren't surveyed, so a ticket raised by an office manager on behalf of six people produces exactly one opinion.
When it fires, and how to change that
Delivery isn't hardcoded. It runs through a system automation, the one that requests a satisfaction rating a set number of hours after the ticket is solved. It is an ordinary automation, which means you can open it and edit the conditions like any other.
That's the single most underused fact about Zendesk CSAT. Most teams treat the survey as a fixed feature and never look inside it.
Conditions worth adding once you're in there:
closed_by_merge was closed administratively. Surveying the customer about it means asking them to rate a ticket that never got an answer.Timing is the other lever. Too fast and you catch people before they have tried the fix. Too slow and you catch people rating their memory. A few hours suits simple tickets, the next morning suits anything the customer has to verify. Pick one and leave it alone, because a score that moves when you change the delay is measuring your automation.
Response rates, and what moves them
Expect a minority to answer. Support surveys behave like every other unsolicited email, and the honest planning assumption is that most solved tickets produce no rating at all.
The things that genuinely lift the rate are unglamorous. Put the rating buttons in the email body so answering is one click rather than a page load. Check how the mail renders on a phone, because that's where it'll be opened. Keep the subject plain and specific to the ticket. And send from the address the customer has been talking to, not a no-reply.
What doesn't lift it: asking more questions, adding branding, or sending a reminder. Reminders on support surveys tend to annoy the people least likely to answer anyway.
Reading the result without fooling yourself
Your satisfaction score is good ratings divided by all ratings received. It isn't out of everyone you helped. That distinction gets lost in every meeting where somebody says "94% of our customers are happy".
Three habits keep the number useful.
Segment before you conclude. By channel, by group, by first response band. An aggregate score hides the one queue that's quietly failing.
Turn on satisfaction reasons. A dropdown asking why a rating was bad turns an unusable number into a diagnosis, and it costs one configuration change.
Watch the sample size. At thirty responses a month, three or four bad ratings swing the headline by double digits. Run the numbers through the CSAT calculator before you hold a meeting about a move that's just noise.
For the wider argument about what the score is worth at all, the customer satisfaction guide goes further.
Where the results show up in reporting
Satisfaction is available in the Support Tickets dataset in Explore, and the attribute you want distinguishes offered, unoffered, good and bad. Getting the denominator right is most of the work here.
A second habit worth building: look at the ratings that arrived without a comment separately from the ones that came with one. A bad rating with no explanation is a data point. A bad rating with three sentences attached is a piece of work you can actually do, and it deserves to be routed to a person rather than counted in an average.
Report on rated tickets when you want the score. Report on offered versus rated when you want the response rate, which is the number that tells you whether the score means anything. Teams almost always build the first and never the second.
One filter to apply everywhere: drop tickets tagged closed_by_merge. If duplicates get merged after a survey has already gone out, the ratings attached to them are noise from a conversation that was folded into another one.
Frequently asked questions
What does Zendesk satisfaction survey setup involve?
Enable the Zendesk CSAT survey, choose when it fires relative to solved, and edit the wording. The Zendesk satisfaction rating is good or bad, and the reasons list underneath is the part worth configuring properly.
How do I turn on the Zendesk satisfaction survey?
Enable satisfaction ratings in the admin settings for Support. That switches on the system automation that emails the requester after a ticket is solved.
Can I change when the Zendesk survey is sent?
Yes. Delivery runs through an ordinary automation, so you can edit the hours-since-solved condition and add exclusions like any other automation.
Do CCs get the satisfaction survey?
No. Only the requester is surveyed, which is worth remembering on tickets raised on behalf of a group.
What is a good Zendesk CSAT response rate?
Lower than most people expect, and it varies hugely by audience and channel. Track your own offered-versus-rated ratio over time rather than chasing a published benchmark.
Should merged tickets be surveyed?
No. Add a condition excluding the closed_by_merge tag so customers are not asked to rate a ticket that was closed administratively.
Two tickets, two surveys, one customer
Ticket Merger catches the second ticket before it becomes a second survey invitation for the same conversation.
Start free trial14-day free trial. No credit card required.