01
Low scores get diagnosed, not logged
A 2-out-of-5 with an empty comment box is a lost customer and a mystery. The AI asks what went wrong in the customer's own terms, immediately.
CSAT questions that earn their rating.
The CSAT question is one line and a scale, the hard part is what you ask next. Below: correct core wording for each touchpoint, and the follow-up sets that turn a 3-out-of-5 into something a team can fix. Diaform asks them adaptively, in a real conversation.
14-day free trial · No demo required
Why CSAT follow-ups should be a conversation
02
Follow-ups adapt to the answer
"The agent was slow" and "the answer was wrong" trigger different next questions. Static forms treat them identically; a conversation doesn't.
03
The same scale, more signal
Your CSAT trend stays intact, the rating question is unchanged. What changes is that every rating arrives with its reason summarized.
04
Recovery inside the conversation
When a fixable complaint appears, the AI can share a help link, offer a gesture, or book a call with support, configured by you, fired in the moment.
05
Sentiment separates identical scores
A resigned 4 and a delighted 4 are different customers. Sentiment on every transcript tells them apart, which the number alone never will.
06
Themes ranked across responses
Aggregate analytics turn hundreds of follow-up conversations into the ranked list of what's actually dragging your CSAT down.
CSAT question sets by touchpoint
Copy these directly, or run them in Diaform, where the AI asks the follow-up automatically after every rating.
01
The core CSAT question (adapt the bracket)
The standard wording. Keep the scale consistent across touchpoints so scores stay comparable.
MESSAGE
How satisfied were you with [your support experience / your purchase / your onboarding]? Scale: 1 (very dissatisfied) to 5 (very satisfied) CSAT = % of responses rating 4 or 5
02
Post-support CSAT
Send within an hour of ticket resolution, while the interaction is fresh.
MESSAGE
1. How satisfied were you with the support you received today? (1-5) 2. What about the interaction most influenced your rating? 3. Was your issue fully resolved? If not, what's still open? 4. How was the speed of the response, and the quality of the answer, separately? 5. What should we do differently next time?
03
Post-purchase CSAT
Trigger right after checkout or delivery.
MESSAGE
1. How satisfied are you with your purchase? (1-5) 2. What almost stopped you from buying? 3. Did anything about the experience surprise you, good or bad? 4. How did the product compare to what you expected? 5. What would make the next purchase easier?
04
Onboarding CSAT
Send after onboarding completes, or at day 14.
MESSAGE
1. How satisfied are you with your first weeks using [product]? (1-5) 2. What were you hoping to achieve, and how far have you gotten? 3. Where did you get stuck or slowed down? 4. What helped you most in getting started? 5. What's the one thing we should improve for new customers?
05
Low-score follow-up (ratings 1-3)
The conversation that matters most. Ask immediately, while recovery is still possible.
MESSAGE
1. What specifically fell short for you today? 2. What were you trying to accomplish when it happened? 3. What would have turned this into a 5? 4. Has this happened before, or was today unusual? 5. Is there something we can do right now to put it right?
Frequently asked questions.
The practical details behind setting up, running, and scaling this kind of research with Diaform.
The standard is "How satisfied were you with [specific interaction]?" on a 1-5 scale. The two rules: name a specific, recent experience (not the company in general), and keep wording and scale identical over time so the trend is real. The value then comes from the follow-up questions, not from rewording the rating.
More conversations your team can run.
Explore the other ways Diaform can turn a static request for feedback into a useful conversation.
Your customers have more to say.
Get an AI feedback agent that listens, follows up, and turns every conversation into answers your team can act on.
Launch your first AI feedback agent in minutes.