Guide6 min readUpdated
Survey Design: How to Write Customer Survey Questions
Survey design for small teams: how to write customer survey questions, pick rating scales and run a survey on WhatsApp or Google Forms.
By Tithi Mondal, UI/UX Designer
A bakery owner in Siliguri sends a 25 question feedback form to her customers. Four people answer. Two of them leave half the form empty. She learns almost nothing and decides surveys do not work. The problem was not surveys. The problem was the design of the survey.
Survey design means choosing what to ask, how to word each question, in what order, and where to send it, so that real customers finish it and the answers help you decide something. A survey tells you what many people think. An interview tells you why. If you need the why first, start with user research for startups and come back here.
What should a survey help you decide?
A survey should help you make one decision, and you should write that decision before you write any question. For example: "Should we add Bengali to the app?" or "Which of these three delivery slots should we keep?"
Write the decision at the top of your page. Then test every question with one sentence: "If the answer is A, I will do this. If the answer is B, I will do that." If both answers lead to the same action, delete the question.
This one habit removes most bad questions. It removes the "nice to know" questions such as age, city and favourite colour that make a form long and give you nothing to act on.
How long should a customer survey be?
A customer survey should take about 3 minutes to finish on a phone. In practice that is 5 to 8 questions with mostly tap answers and at most one or two typed answers.
Most of your customers will open the form on a mid-range Android phone, often outdoors or in a queue. Typing long text on that screen is hard, so every typed question costs you responses. Put the easy tap questions first, and the typed question last.
Tell people the length before they start. "5 questions, about 2 minutes" is a promise that makes people begin. A progress bar or "Question 3 of 6" helps them finish. For layout of the form itself, see form design best practices.
Which customer survey questions work best?
The best customer survey questions ask about a real, recent experience and offer clear answer choices. Here is a set a small shop or app can adapt.
- Recent experience: "How did you order from us the last time?" (choices: app, WhatsApp, phone call, in person)
- Success: "Did you get what you wanted?" (choices: yes, partly, no)
- Problem: "If something went wrong, what was it?" (choices plus "other")
- Effort: "How easy was it to finish your order?" (rating scale)
- Language: "Which language do you prefer for messages from us?" (Bengali, Hindi, English, other)
- One open question: "What is one thing we should change?"
Ask about the last time, not "in general". "How often do you order?" gets a guess. "When did you last order?" gets a fact.
Do not ask two things in one question. "Was the delivery fast and the food hot?" cannot be answered if the food was hot but late. Split it into two questions.
How do you design rating scale questions?
A rating scale question asks the person to pick a point on a line, such as from very difficult to very easy. Use 5 points, label every point with words, and keep the direction the same through the whole survey.
| Good scale | Weak scale |
|---|---|
| Very difficult, Difficult, Okay, Easy, Very easy | 1 to 10 with only the ends labelled |
| Same direction in every question | Bad on the left in one question, bad on the right in the next |
| One idea per question | "Rate our price, quality and service" |
Words matter more than numbers. People in different cities and languages read "7 out of 10" differently, but they all understand "Easy". If your customers read Bengali or Hindi more comfortably, write the scale labels in that language and check the words with two customers before you send.
Add a "Not applicable" choice when some people may not have used the feature. Otherwise they will pick the middle point, and the middle point will then lie to you.
Should you run a survey on WhatsApp or Google Forms?
Run the survey where your customers already read messages. For most Indian small businesses that is WhatsApp. A form link is the most reliable way to collect clean answers, and a short chat is the easiest way to get a reply.
| Method | Good for | Watch out for |
|---|---|---|
| Form link shared on WhatsApp | 5 to 8 questions, easy counting | Link looks like spam from an unknown number |
| Questions asked in a WhatsApp chat | 1 to 3 questions, close customers | You must copy answers into a sheet by hand |
| Printed card with a QR code | Shops, clinics, restaurants | Needs a person to hand it over |
| In the app after a task | Questions about a specific screen | Do not interrupt a payment |
Send the message from a number your customers know, and begin with the person's name and why you are asking. Say what you will do with the answers. Do not send it to customers who have never ordered from you. Respect people who say no, and send reminders rarely. For the message format, see notification UX for push, SMS and WhatsApp.
Google Forms and similar form tools are fine for a first survey, since they are common and let you change questions easily. Check the current options of any tool before you rely on one feature.
How do you read the results without fooling yourself?
Read the results by counting answers, then by reading every typed comment yourself. Do not average rating scale answers across different groups of people.
Start with counts: how many said yes, partly and no. Then split the counts by group, such as new customers and old customers, or app users and WhatsApp users. A mixed total can hide that one group is happy and another is not.
Then read each open answer and move it into a theme in a sheet: "late delivery", "language", "price", "app crashes". The themes that repeat are your list of changes.
Remember what a survey cannot do. People who answer are often the ones who feel strongly, so the silent middle is missing. A survey also says what, not why. When one answer surprises you, call two of those customers and ask for the story, with the story questions from the interview approach.
When you have two designs and want to know which one works better, a survey is not the right tool, and what is A/B testing in design explains the better route.
If you would like a feedback form or survey screen designed to match your app or website, you can get in touch and share the decision you are trying to make.
Frequently asked questions
How many questions should a customer survey have?
Keep it to 5 to 8 questions, about 3 minutes on a phone. Every extra question lowers the number of people who finish.
Is a Google Form good enough for a customer survey?
Yes, for a first survey. It is easy to share on WhatsApp and easy to change. The quality of your questions matters far more than the tool.
Should I offer a reward for answering a survey?
A small thank-you is fine, such as a discount on the next order. Say clearly that the reward does not depend on giving good feedback, or people will write what you want to hear.
What is the difference between a survey and an interview?
A survey collects short answers from many people and shows what is common. An interview is a long conversation with one person and shows why they behave as they do. Most teams need both.
Working on something like this?
Tell Tithi what you are planning. You get a clear scope, a fixed price and a timeline before any work begins. I reply to every enquiry within 24 hours.