Customer satisfaction survey: how to build one, with example questions

What a satisfaction survey is for

To learn what customers won't tell you in person. Most customers who had a bad experience don't complain: they just don't come back. A short survey gives you three things: a number to track whether you're improving, the detail of what's failing (the food, the wait, the service) and, most valuable, the chance to reach an unhappy customer before you lose them.

The rules that get people to answer

• Three to five questions. Every extra question makes more people quit.
• Most important question first. If they only answer one, make it that one.
• Answerable on a phone without zooming or creating an account.
• Right after the experience: at the table, at checkout or the same day. A week later nobody remembers.
• One open question, at the end and optional: "What could we improve?"
• Optional contact details. Make them required and fewer people answer. If they leave them, you can follow up.

Question types and when to use each

To understand the recommendation question and how its score is calculated, read the NPS guide.

Example: restaurant survey

1. From 0 to 10, how likely are you to recommend us to a friend?
2. How was the food? (1 to 5 stars)
3. How was the service? (1 to 5 stars)
4. Was the wait... Quick / Normal / Too long
5. What could we improve? (optional)

Example: auto shops and dealerships

1. From 0 to 10, how likely are you to recommend our service center?
2. Did we return your vehicle on the promised date? Yes / No
3. Did your advisor clearly explain the work and the cost? (1 to 5 stars)
4. Was the problem fixed? Yes / No / Not sure yet
5. What could we improve? (optional)

For service centers, send it the same day as the pickup, by WhatsApp or with a short call, because the customer is no longer on site.

Example: stores and retail

1. From 0 to 10, how likely are you to recommend us?
2. Did you find what you were looking for? Yes / Partly / No
3. How would you rate our staff's service? (1 to 5 stars)
4. What could we improve? (optional)

How to get the survey to customers

• QR code on the table, the receipt or the register. It's the cheapest option and works very well in restaurants. See the QR code survey guide.
• Link by WhatsApp or email after the visit, when you have the customer's contact (service centers, appointments).
• A short call from your team, for high-value customers like a car dealership's.
• A tablet on site, with the same survey open in the browser.

A tip: if you use QR codes across tables or locations, make each code identify where it comes from. That way you know which location, shift or server gets the lowest ratings, without asking the customer anything extra.

What to do with the answers

A survey nobody reads is worse than no survey: you asked for their opinion and nothing happened. At a minimum:

1. Respond to bad ones the same day. Call or message the customer who rated you poorly. Many come back just because someone listened.
2. Check your main number every week and compare it with last month.
3. Look for patterns in open answers: if five people mention the wait, the problem is the wait.
4. Share results with your team, including the good answers with the name of whoever served them.

Is Google Forms enough?

To start, yes: it's free and easy. It falls short when you need to know about a bad rating the moment it happens, compare locations, see trends without building spreadsheets, or catch an unhappy customer before they leave a one-star review.

How SnapRate does it

With SnapRate you build surveys with recommendation questions (0 to 10), star ratings, multiple choice and open text, and share them by QR or link. You generate one QR per table, location or anything you want to measure, and every response is tagged. You decide which answer is an alert (for example, a rating of 3 or lower) and your team gets a WhatsApp message right then. Results show up on a real-time dashboard by location, with daily NPS and CSAT, and you get automatic email reports.

More on customer experience measurement.