What should we decide before we ask employees anything?
Before you ask employees anything, decide what you're trying to understand, what change the answer will serve, and what you'll realistically do with the result. We don't believe organizations need more data for the sake of simply having more data. They need the right data to make better decisions, and that means starting with the decision or problem rather than the survey.
Instead of asking "what should we put on an employee survey?", we're more likely to ask:
- What are you trying to understand? Turnover on one shift and morale across the organization are different questions.
- What changes are you weighing? If the answer won't inform a change, you're collecting data that may never get used, which only hurts morale.
- What do you already know, and what are you assuming? Look at the evidence you already have and what it's telling you, and identify where you're filling in the gaps with assumptions.
- Whose perspective is missing? Different people experience the organization differently. Supervisors, employees, and people who have left may see problems that leadership can't see from where they sit.
- What could you realistically change based on what you learn? Asking about things you can't change sets up disappointment.
Only then should an organization decide what data it needs. If the problem in front of you is turnover, why good people leave despite competitive pay has its own page: Why do good employees keep leaving when we pay competitively?. The full sequence for a larger workforce decision is its own page: We're about to make a big workforce decision. What should we know first?
Why do employees tell the truth in some forms of feedback and not others?
Employees tell the truth when they believe the answer can't be traced to them, when they trust the person asking, and when they believe their feedback will actually matter. Candor is a condition you design, and past experience shapes it. If employees have spoken up before and nothing changed (or worse, there were negative consequences), they learn that being candid isn't worth the risk. What has happened when employees shared feedback in the past can influence how honestly they respond today.
The conditions that raise candor
- Anonymity that's real, and visibly so. A "confidential" survey with a five-person department and a question about job title isn't confidential. Survey researchers treat confidentiality as a design obligation; the American Association for Public Opinion Research lists protecting respondents among its best practices.
- An impartial facilitator, interviewer, or evaluator. People answer differently for a trusted outside party, for HR, and for their own supervisor, so choose deliberately.
- A track record. When employees repeatedly provide feedback and nothing happens, trust is eroded. If that has happened in your organization, acknowledge it before asking again and explain what will be different this time.
- The right group composition. Focus groups of peers, without the supervisor in the room, surface things a survey never will.
If you've already run a survey and nothing came of it, that has its own page: We ran an engagement survey and nothing changed. What went wrong?
Is a survey the right tool, or should we talk to people?
A survey is the right tool when you need to hear from many people about a defined set of questions. Talking to people is the right tool when you need to understand why a pattern is occurring and what it feels like from the inside. Interviews and focus groups can also help you figure out what belongs on a survey in the first place, especially when you don't yet know enough to ask the right questions. Most workforce questions need both, because numbers can show patterns, magnitude, and differences between groups, while conversations explain experiences, context, barriers, and motivations.
| Method | What it's good for | What it can't do alone |
|---|
| Your existing records (turnover, overtime, schedules, complaints) | Showing where the problem is concentrated, at no cost to employees' time | Explaining the reasons |
| A survey | Reaching everyone; comparing groups; tracking change over time | Explaining an unexpected result |
| Interviews | Depth, nuance, and the things people won't write down; informing survey development | Breadth, because they take considerable time |
| Focus groups | Hearing how a team talks about a problem together; informing survey development | Sensitive topics, or any room with a boss in it |
| Supervisor conversations | The view from the middle, which executives rarely get | Employees' own account |
Surveys, interviews, focus groups, administrative data, and organizational metrics become much more powerful when they're used together. Interviews or focus groups might uncover an issue you hadn't considered, which can then be tested across the workforce in a survey. A survey might tell you that one department reports lower satisfaction, and follow-up conversations can help explain why. The method should follow the question, and the least expensive first move is almost always examining the information you already have.
How do I know whether our survey questions are any good?
You can tell whether your survey questions are any good by checking that each one is neutral, asks about one thing at a time, and covers something you're prepared to act on. An employee survey is only as useful as the questions it asks. Poorly designed questions can produce misleading results, miss what employees actually want to tell you, or leave you with data that doesn't help you decide what to do next. Good survey design makes sure you're asking the right questions, in the right way, so you can trust what you learn and use it. Survey design is a discipline with well-documented failure points, and many employee surveys stumble on several of them.
The failures we see most
- Leading questions. "How satisfied are you with our excellent new schedule?" measures politeness. Pew Research Center's guidance on writing survey questions treats neutral wording as the first rule of the craft.
- Two questions in one. "Do you feel supported by your supervisor and your team?" can't be answered by someone who feels supported by one and not the other.
- Length. Every additional measure creates work for the people answering, and fatigue appears as straight-line answers halfway down the page.
- Asking what you won't change. A question about pay when pay isn't on the table sets up a broken promise.
- No plan for the results. The survey closes, a summary circulates, nothing changes. That's the most common failure and the most expensive one, because it teaches employees not to bother next time.
A short survey with five well-built questions, a clear purpose behind it, and a date by which people will hear what was learned usually tells you more than a fifty-item engagement questionnaire.
The Grantibly perspective
Lauren's perspective
I taught research methods to university students for more than a decade, and the lesson that transfers most directly to employers is the simplest one: start with the question, not the instrument. Organizations reach for a survey the way people reach for a thermometer, as if the tool were the diagnosis, and it isn't. Data doesn't make decisions. People do. Our job is to help you turn information into understanding and action.
So when a leader tells us they need an employee survey, we usually ask what they're trying to find out, what they already know, and what they'd be willing to change. Sometimes the answer is a survey. Sometimes it's an afternoon with your own turnover records, a few conversations with supervisors, or a focus group run by someone employees don't report to, and often it's a combination.
And whatever you ask, plan for the answer before you ask it, because employees who tell you the truth are handing you something special. What you do next determines whether they'll ever do it again.
