KoltrixExit Survey Emails: Questions That Get Honest Answers
A short exit survey by email can tell you more about churn than any dashboard. Which questions to ask, how to format them, and what to do with the answers.
On this page(9 sections)
- Why email, and why not a long form
- The two questions that matter
- 1. One multiple-choice reason
- 2. One open question
- Reply-to-answer beats a survey link
- Timing and sender
- Questions to avoid
- What to do with the answers
- Tag and count
- Read the open answers in a batch
- Close the loop
- Share it widely
- A note on response rates
- Exit survey checklist
- Key takeaways
Usage data tells you that a customer left. Only the customer can tell you why. An exit survey email is the cheapest way to find out, as long as it's short enough that people actually answer it and honest enough that they tell you the truth.
Why email, and why not a long form
Most cancel flows include some kind of survey. Many of them are long forms with a dozen fields, half of them required, standing between the customer and the cancel button. People click whatever gets them out fastest, and you end up with data that's mostly noise.
An email sent after cancellation works differently. The decision is already made, so there's no pressure, and nobody feels they're being blocked. People who reply are telling you something because they want to. You get fewer responses than a forced form, but the responses mean something.
The best setups use both: one optional question in the cancel flow, and a short follow-up email a day or two later.
The two questions that matter
You can learn most of what you need from two questions.
1. One multiple-choice reason
Give people a short list of the reasons you actually see, plus "Other." Keep it to six or seven options at most:
- It was too expensive for what we used
- It was missing something we needed
- We had trouble getting it set up
- We switched to a different tool
- Our project or need ended
- Something went wrong with support or billing
- Other
Write the options in plain language, the way a customer would say them, not in your internal categories. "Lack of perceived value" is your label; "too expensive for what we used" is what they'd actually say.
2. One open question
After the multiple choice, ask one open question. The wording matters more than you'd expect:
- "What would have had to be different for you to stay?" invites specifics.
- "Any other feedback?" invites nothing.
- "What did you switch to, and what does it do better?" is direct, and many people answer it.
Pick one. Two open questions halve your answers to both.
Reply-to-answer beats a survey link
The easiest format is often no form at all. Write the email so people can reply with a number or a sentence.
Subject: One question before you go
Hi Priya,
Your Pinecraft subscription has been cancelled, and your account
stays readable until June 30.
If you have ten seconds, could you hit reply and tell me which of
these was closest?
1. Too expensive for what we used
2. Missing something we needed
3. Hard to get set up
4. Switched to another tool
5. Our need ended
6. Something else
Just the number is plenty. If you want to say more, I read every
reply myself.
JordanPinecraft is hypothetical, but notice what the format does. It costs the customer one keystroke, it shows up in their normal inbox, and the "I read every reply" line makes it a conversation instead of a data-collection exercise. It only works if the reply-to address goes to a person who actually reads it. A noreply@ sender kills this entirely.
If you need structured data at scale, a one-question survey link with the options as clickable buttons is a reasonable middle ground. Each button records the answer on click and opens a page with the optional open question.
Timing and sender
- When: within one or two days of cancellation, while the experience is fresh. Not the same minute as the cancellation confirmation; that email has its own job (see our guide to the cancellation confirmation email).
- From whom: a named person. For an early-stage company, the founder. For a larger one, whoever owned the relationship.
- How often: once. No reminder. Someone who doesn't answer an exit survey has answered it.
Questions to avoid
| Question | Problem |
|---|---|
| "On a scale of 1 to 10, how likely are you to recommend us?" | They just left. The number tells you little and feels tone-deaf. |
| "Why did you decide to leave us?" | Puts them on the defensive. Ask what would have had to be different instead. |
| "Would you consider coming back if we offered a discount?" | That's a sales pitch dressed as a survey, and people can tell. |
| Required demographic or company-size fields | You already have this data, or should. |
| Anything with more than one open text box | Response rates drop with every extra box. |
What to do with the answers
Exit survey data is only useful if somebody looks at it regularly and acts on it.
Tag and count
Every reply gets one reason tag. Keep a simple running tally by month. You're looking for shifts: if "hard to get set up" doubles after a release, something in that release made onboarding worse.
Read the open answers in a batch
Once a month, read every open answer in one sitting. Patterns jump out when you read thirty in a row that you'd never notice reading one a day. Pull exact phrases into your product notes; customers' own words are better than paraphrase.
Close the loop
When you fix something people left over, tell them. "You mentioned the Slack integration was missing when you left. It shipped today." That's the most effective win-back email there is, and it's only possible because you asked.
Share it widely
Exit survey results shouldn't live only with the founder or the customer success lead. Engineers and designers benefit from seeing, unfiltered, why paying customers walked away.
A note on response rates
You won't hear from everyone, and that's fine. Resist inventing a benchmark to judge yourself against. What matters is whether you're getting enough answers each month to see patterns, and whether the answers are specific enough to act on. If replies are rare, check the basics: is the email arriving in the primary inbox, is it from a real person, and is it short enough to answer on a phone?
Also watch for skew. People who had a strong negative experience are more likely to reply than people who drifted away. Weigh the answers against your usage data before drawing big conclusions.
Exit survey checklist
- One optional question in the cancel flow, never required
- Follow-up email within one or two days, from a named person
- One multiple-choice reason in customer language, plus "Other"
- One open question, worded to invite specifics
- Reply-to address that reaches a human
- No reminder, no discount pitch
- Monthly tally of reasons and a batch read of open answers
- A habit of telling people when you've fixed what they left over
Key takeaways
- Ask after the cancellation, by email, when there's no pressure on either side.
- Two questions are enough: one multiple-choice reason and one well-worded open question.
- Let people answer by replying, and make sure a person reads the replies.
- Tag, count and read answers on a schedule, and look for changes over time.
- When you fix a reason people left, tell them.
Start with Koltrix
Your domain, one inbox, and an API that sends.
A team inbox where AI sorts and drafts (nothing is sent without your click), plus the transactional API and SMTP relay your product sends with. 7 days free, no card.
