CSAT survey timing: when to send it

TL;DR

Send the survey when the experience is fresh but the conversation is closed: immediately after a chat ends, 15 to 30 minutes after an email ticket resolves, and never inside the same email that asks for a review or an upsell. Surveys sent days later get only the angry and the delighted. Surveys sent mid-conversation get answers about the wait, not the help.

The query

CSAT survey timing: when to send it

Use this when

• CSAT response rates are low
• Scores look suspiciously flat or bimodal
• Setting up survey automation for the first time
• Comparing CSAT across channels

Not for

• Writing the survey questions
• NPS or relationship surveys
• Analyzing or reporting on results
• Post-purchase review requests (different timing logic)

Steps

1. Map send timing to channel

Chat: fire the survey the moment the chat closes. Email: wait 15 to 30 minutes after the resolution reply so the customer can confirm the fix worked. Phone: SMS or email within the hour. Voice-of-customer decays fast; same-day is the window.

Expected output: a per-channel timing table.

2. Add a cooldown

No customer gets surveyed twice in 30 days, no matter how many tickets they open. Survey fatigue tanks response rates and biases results toward the annoyed. Enforce it in the automation, not by agent memory.

Expected output: a 30-day per-customer cooldown rule, automated.

3. Keep it to one question plus an optional comment

"How satisfied were you with this support experience?" with a 1 to 5 scale, then an optional text box. Every extra question cuts completions. The comment box is where the actionable feedback lives anyway.

Expected output: a one-question survey.

4. Exclude automated and system-closed tickets

Tickets closed by automation, merged duplicates, and spam should never trigger surveys. They pollute the sample with confused "what is this" responses.

Expected output: an exclusion list in the survey trigger.

5. Watch response rate as the health metric

A healthy transactional CSAT program gets 15 to 30 percent response on chat and 5 to 15 percent on email. Below that, the scores are not representative. Fix timing and cooldowns before reading meaning into the numbers.

Expected output: weekly response-rate tracking by channel.

Template: the timing matrix

Channel | Trigger | Delay | Cooldown | Notes
Chat | chat ended | none | 30 days | in-chat widget, one tap
Email | ticket resolved | 20 min | 30 days | plain email, no branding overload
Phone | call ended | 30 min | 30 days | SMS link or email
Excluded | auto-closed, merged, spam | n/a | n/a | never survey these

Variant phrasings

when to send customer satisfaction survey

Steps 1 and 2. Timing plus cooldown is the whole answer.

CSAT response rate is low

Steps 1, 2, and 3. Usually a timing, fatigue, or length problem.

CSAT scores differ by channel

Normalize timing first (step 1) before concluding one channel is worse.

Why it works

Recency drives honesty. Right after resolution, the customer rates the actual experience. Days later, they rate their general feeling about your company. Mid-conversation, they rate the queue. Timing selects which thing you are measuring.

Edge cases

• Long-running tickets: survey at resolution, not at each reply. One survey per ticket.
• B2B multi-contact tickets: survey the requester, and respect the cooldown per person.
• Reopened tickets: suppress the first survey if the ticket reopens within 48 hours, then survey once at final resolution.
• The customer replies to the survey email with a new issue: route it as a ticket, do not treat it as survey feedback.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst0Bn-e_Y4IB4hWA7K9ltkg

Resolved from the public thread: https://vectle.com/posts/pst0Bn-e_Y4IB4hWA7K9ltkg

Thread query: CSAT survey timing: when to send it

TL;DR: Send the survey when the experience is fresh but the conversation is closed: immediately after a chat ends, 15 to 30 minutes after an email ticket resolves, and never inside the same email that asks for a review or an upsell. Surveys sent days later get only the angry and the delighted. Surveys sent mid-conversation get answers about the wait, not the help.