Does your company issue customer surveys, but you’re not getting the insights you need? Read on for a quick class on how to write good survey questions that result in objective, actionable data.
Are you looking to improve your data right away and don’t want to read this entire post? Here are three tips I always give:
- Ask compelling but non-leading questions
- Craft an engaging survey invite—because getting a high response rate is key
- Combine your surveys with other methods
Sit Back and Watch the 1-minute Video
Interaction Metrics’ Kaitlyn Bartley gives you an example of a leading question and a power word to use in your survey invite. She also asks whether in fact, you need a customer survey!
Before I get on with how to write good survey questions, let’s level set on why you are writing a customer survey in the first place.
Getting Clear on Your Survey ‘Whys’
The three main reasons to do a customer survey are straightforward: gauge your customers’ pleasure/displeasure, discover process problems you may not know about, and pinpoint opportunities to improve. That’s it. That’s the purpose of your customer survey.
Unfortunately, many surveys that start as a vehicle for measuring customers’ perceptions wind up going down countless paths. This is tiring for your customers and loses actionability for you.
Instead, keep your survey questions focused. And if you have multiple objectives and a large audience, you probably need several customer surveys, each with its own goal.
The key to understanding your customers’ thoughts and feelings lies in the questions you ask.
Even if you use the Interaction Metrics GPT as a guide, the best survey, like the best writing, will be customized to your audience, your company, and your goal.
Sure, you can copy questions from another company’s survey.
However, this approach will fail to provide data about the unique interactions that take place between your customers and you. So, your first task is to write incisive survey questions–or hire a writer to do it for you.
Tips for How to Write Good Survey Questions
First of all, recognize that it’s easy to introduce flaws into any survey.
How do you know when your survey is flawed?
If you’re changing your business strategy based on survey data and yet you’re failing to get effective results, then you probably have a flawed survey.
Onward with eight tips to get you going in a good direction with sound survey design.
Let Your Customers Answer Anonymously
One of your initial questions should ask respondents if they would like to answer your survey anonymously. Roughly half of all respondents choose to take surveys anonymously, and everyone’s feedback is valuable, even if it’s anonymous.
Scour Your Survey to Prevent Leading Constructs
A survey is a flop if it assumes your customers already feel a certain way about your company or products.
For example, don’t ask how satisfied your customers are with aspects of your company because that assumes they are somewhat satisfied.
Instead, ask how your customers would rate your company attributes. This is a more neutral approach, which matters because a good survey gives your customers’ honest opinions.
Once you’ve written your survey, test it with users. Find out if your questions allowed them to let their hair down and express themselves candidly. BTW, biases can be subtle, so test and edit your questions several times.
Keep Your Questions Centered on a Single Objective
A survey is not the place to explore a zillion different topics. Keep your questions limited to one subject. This way, your survey will be short and simple. In turn, you’ll have a high completion rate, which translates into more data for you.
Add Incentives to Boost Response Rate
To boost your survey response rate, add incentives like a gift card, charity donation, or priority code. The fact is that to ensure your data is representative, in almost all cases, you need a high response rate.
For information about statistically valid data with a sample size calculator, go here
But how do you get customers to take your survey in the first place? It starts with an SMS or email invite. In your invite, use persuasion principles like social cues and power words to encourage customers to click on your link.
Use Language that Reflects the Way People Speak
When was the last time someone told you they “neither agreed nor disagreed” with you? It’s probably never happened because people don’t talk like that.
Whenever possible, write survey questions that sound natural. For example, ratings scales that use words like “Bad,” “Poor,” “OK,” “Good,” and “Great” work because these are words we’re all familiar with and use every day.
Nix Double-Barreled Questions
These questions ask about two separate issues in one question. For example, “Was your waiter prompt and polite?” asks about two behaviors in one sentence. This muddles your data and obscures the information you need to improve the customer experience.
Don’t Ask Customers Irrelevant Questions
If your customer isn’t a gym rat, don’t ask them how often they go to the gym. Allow your respondents to opt out of questions that aren’t pertinent. Even better, use logic gating to make sure customers only see questions that are germane to them.
Consider Scrapping Your Survey Altogether
Surveys have become so ubiquitous that you might get more data by conducting customer interviews, examining contact center conversations, or asking questions by email without linking to a survey or a form. Even better, use a mixture of research methods because each method will result in different kinds of actionable insights.
Survey Design Encapsulated
Good survey design starts with knowing how to write good survey questions.
But then it’s also about getting enough responses to make sure your data is reliable.
And the best survey design always starts by asking what kind of survey you really need or whether customer interviews, usability studies, or observational studies might serve you even better.
================================================
Want accurate data and an actionable survey? Get in touch.
================================================
