Key takeaways
- Ask at moments when the experience is fresh and the customer has a reason to answer, such as right after a support chat, a delivery or a first week of use. Random quarterly blasts get the lowest-quality answers.
- Keep each request to one main question plus an optional comment. A short survey that people finish beats a thorough one they abandon.
- Set a contact limit per customer, for example one survey every 30 to 60 days across all channels, so feedback requests never pile up.
- Closing the loop is what makes people answer next time. Reply to negative answers personally and tell customers what changed because of their feedback.
- Survicate is our top pick for collecting feedback across email, website and product. Tidio is the simplest way to capture feedback inside support conversations, where many of the most useful comments start.
Every customer has received the survey that asks fifteen questions about a two-minute purchase, then a reminder, then a second reminder. Most of them close it. The few who answer are often the most annoyed or the most bored, which is not the sample you want to plan from.
Feedback is still worth collecting. It tells you why people leave, what confuses them and which problems keep coming back. But the way most businesses collect it trains customers to ignore it.
The late-summer lull is a good time to fix this. Volume is lower, and a feedback setup built now will have a few months of answers behind it before the busy autumn season. This guide covers when to ask, how to ask, and how to make sure answers actually change something.
Why feedback requests annoy people
The common mistakes are easy to spot from the customer's side:
- Asking at the wrong moment. A survey about "your experience" three weeks after delivery asks people to remember something they have forgotten.
- Asking too much. Ten questions where one would do. Every extra question lowers completion.
- Asking too often. A CSAT survey, an NPS email and a product poll in the same week, from teams that don't talk to each other.
- Asking the wrong people. Surveying everyone about a feature only some of them use.
- Asking and then ignoring. Customers who report a problem and hear nothing back stop answering.
Fixing these is mostly about planning, not tools.
Step 1: Decide what you need to learn
Start with the decisions you want feedback to inform, then work back to questions. Three questions cover most small businesses:
- Was this interaction good? Measured right after a support conversation or a purchase (CSAT).
- Would they recommend us? Measured occasionally to track the relationship over time (NPS).
- Why did they leave or not buy? Measured at cancellation, at returns or on an abandoned checkout.
If you can't name the decision a question will change, drop the question. For the difference between the three metrics and when to use each, see CSAT, NPS and CES explained.
Step 2: Ask at the right moment
Timing matters more than wording. Ask when the experience is fresh and the customer has a natural pause:
| Moment | What to ask | Best channel |
|---|---|---|
| A support chat or ticket just closed | How did we do? (rating plus optional comment) | Inside the chat, or email shortly after a ticket |
| Order delivered | Did everything arrive as expected? | Email or SMS a day or two after delivery |
| First week of using a product or app | How easy was it to get started? | In-product survey |
| Cancellation or return | What was the main reason? | Inside the cancellation or returns flow |
| Every six to twelve months | How likely are you to recommend us? |
If you have to remind people what they are rating, you asked too late.
Step 3: Keep it to one question
The best feedback requests have one main question and one optional free-text box:
- CSAT: "How would you rate the help you received?" with a five-point scale.
- Effort: "How easy was it to sort this out?"
- Open follow-up: "Anything we could have done better?"
The rating gives you a trend; the comment tells you why. If you need more detail, ask a follow-up only to the people who gave a low score.
Step 4: Use the channel the customer is already in
Asking inside the conversation or product gets far more answers than an email that arrives later. Ratings shown in a chat window as it closes are answered while the customer is still there. In-product surveys catch people while they use the feature.
Email still works for moments that happen offline, like a delivery, and for occasional relationship surveys. Keep emailed surveys short enough that the first question can be answered from the email itself.
Step 5: Set a contact limit
Agree one rule across the company: no customer receives more than one survey in a set period, for example 30 to 60 days, except the rating at the end of a support conversation they started. Most survey tools can enforce this with throttling. Without a rule, marketing, support and product will all ask the same person in the same week.
Step 6: Close the loop
Closing the loop is what separates feedback programmes that work from ones that fade:
- Reply to every low score within a day or two, personally, and try to fix the issue.
- Tag comments by topic so you can count them: delivery, product quality, pricing, a specific feature.
- Review the themes monthly with whoever owns them, and pick one or two to fix.
- Tell customers what changed. A short "you told us, we changed it" note in a newsletter or help center makes people more willing to answer next time.
Turning these themes into product decisions is its own discipline; we cover it in how to turn support conversations into product insights.
Common mistakes
- Treating the score as the goal. Teams that chase a number start nudging customers to rate high. Read the comments instead.
- Surveying only happy moments. Feedback after cancellations and returns is often the most useful.
- Hiding the free-text box. Without comments, a falling score tells you nothing about the cause.
- Running surveys in five tools. Spread-out feedback never gets read. Keep it in one or two places tied to the customer record.
The best tools for collecting customer feedback
We compared tools on the channels they cover, how short and well-timed their surveys can be, how easily answers reach the team that can act on them, and whether a small team can run them without a researcher.
Survicate runs short surveys by email and link, as website pop-ups, inside web products and in mobile apps, and it connects to Zendesk, Intercom and HubSpot so feedback lands next to the customer's history. It suits teams that want one place for post-support CSAT, NPS and product feedback, with throttling so the same person isn't asked twice in a week.
Why we like it
- Covers email, website, in-product and mobile surveys
- Two-way help desk integrations with Zendesk, Intercom and HubSpot
- Good controls for targeting and survey frequency
Watch out for
- No permanent free plan, only a short trial
- More than a very small business needs for one simple CSAT question
Tidio asks for a rating and an optional comment automatically when an agent closes a chat or solves a ticket, using a simple five-point emoji scale you can reword. Because the rating sits on the conversation itself, you can open any low score and read exactly what happened, including conversations handled by Lyro, its AI agent. For a small team, that covers the single most valuable feedback moment without another tool.
Why we like it
- Rating requests sent automatically after chats and tickets
- Every score linked to the full conversation
- Satisfaction tracked for AI and human conversations
- No extra tool or setup for a small team
Watch out for
- The live chat rating works in the Tidio widget, not on Messenger, Instagram or WhatsApp
- Not built for NPS programmes or product surveys; pair it with a survey tool
Typeform shows one question at a time, which makes longer surveys feel lighter and works well for onboarding questionnaires or annual customer surveys. The free plan is enough to test it, though response limits arrive quickly.
Why we like it
- Pleasant one-question-at-a-time format
- Strong logic and many integrations
Watch out for
- Free plan has a small monthly response limit
- Overkill for a one-question CSAT survey
Formbricks runs link, website and in-product surveys, with mobile SDKs, and can be self-hosted if you want feedback data on your own servers. A good fit for product teams with a developer to set it up.
Why we like it
- Open source, with a self-hosting option
- Free plan covers a modest monthly response volume
Watch out for
- In-product surveys need developer setup
- Fewer ready-made help desk workflows than Survicate
Sprig is built for product teams that run continuous in-product studies and want AI to design and summarise them. It is priced through sales and suits larger companies with a research programme.
Why we like it
- AI-assisted study design and analysis
- Strong in-product and mobile targeting
Watch out for
- No free plan or public pricing
- Built for product research, not post-support CSAT
A simple setup for a small business
- Turn on ratings after support conversations in your chat or help desk. In Tidio, that is a toggle in the customer satisfaction settings.
- Add one post-delivery or onboarding survey with a single question, using Survicate, Typeform or Formbricks.
- Add a reason question to your cancellation or returns flow.
- Set a 30 to 60 day contact limit and agree it with everyone who sends surveys.
- Book a monthly half hour to read low scores and comments, and pick one thing to fix.
Comparing survey tools in more detail? See the best survey tools for small businesses or browse the customer feedback category.
Frequently asked questions
How often should you ask customers for feedback?
For relationship surveys such as NPS, once or twice a year per customer is plenty. For transactional surveys such as CSAT after a support chat, asking every time is fine, as long as you cap how many different surveys one person receives in a month across all channels.
What is a good response rate for a customer survey?
It varies by channel. Ratings shown inside a chat that just ended usually get far more answers than emailed surveys, and in-product surveys sit somewhere between. Compare your own rate over time rather than chasing someone else's benchmark.
Should I offer an incentive for completing a survey?
For private feedback surveys, a small thank-you such as a prize draw is fine and can lift response rates, but it can also attract careless answers. Never tie incentives to public reviews or to giving a positive answer.
How do I get feedback from customers who never answer surveys?
Read what they already tell you. Support conversations, cancellation reasons, returns notes and public reviews are feedback too. Tagging support conversations by topic, which tools like Tidio make easy, often reveals more than a survey would.