5 tips for employee surveys that actually make a difference - Work Life by Atlassian (2024)

Surveys are a tool for getting to know people. That makes them a great place to start if you want to build a culture where people feel valued, connected to one another, and inspired by their work.

But effective employee surveys don’t happen by accident. Constructing surveys that return useful results – surveys that actually make peoples’ lives better – is both an art and a rigorous science.

Through my academic research in psychometrics and quantitative psychology, I’ve spent my career studying how we can use analytical methods to understand people’s experiences and behaviors. Now, as a senior researcher on Atlassian’s People Insights team, I help design surveys that meaningfully improve work and life for our employees.

You might not have the resources to work with a professional survey scientist. But by following these best practices for collecting and analyzing human data, you can start gathering the knowledge you need to create a better employee experience.

What are employee surveys?

An employee survey is any assessment designed to capture data about how employees experience their work.

Employee surveys might ask open-ended questions, requiring written responses. Or, they might capture more quantitative data with numeric scales or multiple-choice questions.

Typically, organizations conduct employee surveys because they want to create a better employee experience. They need data to understand both the current baseline they’re starting from, and specific actions they can take to improve it.

Of course, “employee experience” is a broad and all-encompassing term! Good employee surveys often target and investigate specific parts of this experience, like onboarding, job satisfaction, or manager-employee relationships.

The power of effective employee surveys

Obviously, improving peoples’ working lives is a worthwhile goal on its own. But happier, more inspired employees elevate their organization in very tangible ways, too.

Throughout my career, I’ve seen that ultimately, more engaged and happy employees perform better, are more productive, and stay with the organization longer. All these qualities can help a company achieve its goals –and surveys are the most direct way to uncover how you can achieve them.

Another excellent survey goal is to foster belonging –a sense of personal connection to the organization, its values, and the people within it. Belonging is an often-overlooked precursor to engagement – according to some research, 91% of employees who feel they belong are engaged,compared to 20% of employees who don’t.

5 ingredients for a survey that’s worth your time

Want your survey to help you create positive change? These five principles are a great place to start.

Know your goal

Instead of trying to capture every aspect of your employees’ experience, think critically about your goal. What actual outcomes are you trying to achieve? What are you trying to move the needle on?

Then, work backwards from your desired outcome. What aspects of the employee experience impact the outcome you want, and how can you investigate them? What questions will help you take an informed action?

When your survey doesn’t have a clear goal, you may end up asking questions you can’t take action on. That’s a surefire route to survey fatigue, damaged employee trust, and diminished morale.

For example, maybe you want to reduce turnover. So, you could ask questions about employees’ relationships with their managers, desired future prospects with the organization, and work-life balance.

Focus on actionability

Once your goal is clear, it’s easier to ask questions that will help you achieve it. While you can include some general, “sense check” questions, your survey should be geared towards informing concrete action.

Actionability has three components:

• Ask questions that can inform action. For example, “what does an ideal work-life balance look like for you?” is more actionable than “how satisfied are you with your work-life balance?”
• Don’t ask questions you can’t take action on. A classic example is “How satisfied are you with your compensation?” Nearly everyone would like a raise, but it’s unlikely an employer can grant one to their entire workforce at the same time.
• Show people you’re taking action. For example, you could hold an all-hands meeting after your survey, sharing the data and outlining your action plan. Then, check back in periodically to follow up and show them you meant what you said.

Keep context in mind

Every organization is different, and the employee experience changes over time. If you don’t keep organizational climate in mind when designing questions, your survey can come across as insensitive, and return unhelpful answers.

Here are some examples of how context can inform survey questions:

• Asking a question like “How optimistic are you about the future with this company?” after there’s just been a round of layoffs and morale is low will most certainly return unfavorable responses
• If your workforce includes hourly, salaried, and contractor employees, don’t ask all of them the same questions about compensation.
• If you ask people’s opinions of managers or leadership right before their annual performance review, you might get biased responses.

Test out (pilot) your survey items

“Survey piloting” might sound technical, and it definitely can be! Survey scientists use sophisticated statistical modeling and cognitive testing techniques to make sure survey items will actually drive the changes they’re looking for.

But piloting is a great idea for anyone who’s running an employee survey. Piloting survey items essentially means testing them, with a small random sample, before releasing them to your entire organization. Piloting helps improve your questions, so you’ll get better data when you run the survey at full scale.

It’s very important that your sample be randomly selected – don’t just ask your teammates to pilot the survey with you. The idea is to understand how the survey will land with people in different demographic groups, areas of the company, and levels of seniority.
If you work at a small organization, you can still gain meaningful insights from a pilot survey – even a focus group as small as five people is worthwhile. You can also try piloting more questions than you’ll want in the finished survey, so you can identify the ones that perform best.

There are two ways of testing survey questions: qualitatively and quantitatively.

Interpret your results critically

Evaluating your answers is just as important as asking the right questions. When you get your results back, don’t just take them at face value! Even if you carefully tested your survey, there are many ways data can be biased or skewed.

Here are four best practices to follow when interpreting survey data.

Consider response rates

If not many people responded to your survey, you won’t have statistically significant data – and the data you do have might not accurately represent your organization. For example, maybe respondents skewed heavily towards employees in your department, because you’d reminded them multiple times.

If you don’t have enough data, or it’s not distributed across your company evenly, you may need to re-run your survey. To avoid this problem, send the survey to double or triple the number of participants you’ll actually need to get useful data.

Look for irregularities

Normally, survey responses roughly distribute along a bell curve. Answers may trend in one area, but there will be responses on either side.

If you see dramatic trends that don’t follow this distribution, it probably indicates an issue with your question. For example, if almost everyone responded to a question with “strongly agree” or “strongly disagree,” that typically means your item is not worded in a way that would elicit varying opinions.

If everyone shares the same super-strong opinion about something, there’s very little additional information you can glean, or actions you can take from those results.

Use benchmarks whenever possible

Say 70% of your survey respondents love their managers. Is that good or bad? One way to know is to use benchmarks – a company or industry standard to compare your results against.

Some companies compile and sell benchmark data for this exact purpose. Even better, you can use your own company’s past performance as a starting point.

It’s not the end of the world if you don’t have benchmark data, and you shouldn’t over-rely on it anyway. You’ll get better results by focusing on the outcomes that matter to you, not pushing to get engagement from 76% to 80% just because that’s the industry standard.

Focus on what will actually move the needle – not on your “worst” results

One super-common pitfall when interpreting survey data is to focus attention where the company performed worst.

In reality, action should be directed not to what feels like an emergency, but to areas that will have the biggest impact on your goal. In survey science, this is referred to as “predictive power.”

For example, say your desired outcome was improved employee retention, and survey respondents were highly dissatisfied with manager communication. That doesn’t mean you need to push every manager into communication training!

What if a more generous vacation policy would do more to entice your employees to stay? In that case, a question about vacation would have more “predictive power,” and would be a better way to guide action.

Survey scientists use statistical modeling to evaluate predictive power. Even if that’s not within reach for you, it’s a good reminder to stay focused on your goal – not rush into hasty actions because you underperformed in one area.

Special thanks to Genevieve Michaels for her contributions to this article.