Employee Satisfaction Surveys: What to Ask, How to Run One, and What to Do Next

Most employee satisfaction surveys don't fail because of the questions. They fail afterward. People answer honestly, the results go to a slide deck, and nothing visible changes. The next survey gets fewer responses and blander answers, because everyone learned what happens to honest feedback.

A good employee satisfaction survey is a small promise: tell us how it's going, and we'll show you what we did with it. This guide covers the whole loop: what to ask, how to run it without exposing anyone, how to read the results, and how to follow up so the next survey is even more useful than this one.

Satisfaction and engagement are not the same thing

The two words get used interchangeably, but they measure different things, and knowing which one you care about shapes the questions.

• Satisfaction is how people feel about their job and conditions: pay, workload, their manager, flexibility, the tools they're given.
• Engagement is how invested people are in the work itself: whether they care about the outcome, put in discretionary effort, and see a future here.

Someone can be satisfied but disengaged (comfortable, coasting) or engaged but dissatisfied (committed to the mission, burning out). Most small and mid-sized teams want a bit of both, which is why a good survey mixes questions about conditions with one or two about commitment, like the employee Net Promoter question below.

Decide what the survey is for before you write a question

Start with the decision, not the questionnaire. Ask yourself: what would we change depending on the answers?

Good reasons to run one:

• You're losing people and want to know why before more of them leave.
• You changed something big (a reorganization, return-to-office, a new manager structure) and want to know how it landed.
• You want a baseline now so you can measure whether things improve.
• You're about to plan the year and want employee input on where to invest.

If you can't name a decision the results might change, wait. A survey that leads nowhere does more harm than not asking at all.

The questions worth asking

Keep the survey to roughly 8–12 questions. That's about five minutes, short enough that people answer thoughtfully instead of clicking through. Here's a core set, with what each one tells you.

1. Employee Net Promoter Score (eNPS)

How likely are you to recommend this company as a place to work? (0–10)

One question, easy to track over time, and a decent proxy for overall sentiment. Scoring is covered below.

2. Overall job satisfaction

Overall, how satisfied are you with your job? (1–5)

The headline number. On its own it tells you whether something's wrong, not what, which is why the questions after it matter.

3. Feeling valued

How valued do you feel for the work you do? (1–5)

Recognition is one of the cheapest things to improve, and one of the most common reasons people quietly start looking elsewhere.

4. Manager support

How well does your manager support you? (Very well → Very poorly)

People often say they leave managers, not companies. This question is sensitive, so see the anonymity section before you break results down by team.

5. Workload and work-life balance

How satisfied are you with your work-life balance? (1–5)

If this one drops, look at staffing and deadlines before you look at wellness perks.

6. Growth and development

How satisfied are you with your opportunities for growth and development? (1–5)

Especially telling for your strongest people, who usually leave for growth before they leave for pay.

7. Clarity

How clear are you about what's expected of you in your role? (1–5)

Unclear expectations are a quiet source of stress and wasted effort, and they're usually fixable with a conversation rather than a budget.

8. Tools and resources

Do you have the tools and resources you need to do your job well? (1–5)

This one often produces the most concrete, fixable answers.

9. Confidence that feedback matters

How confident are you that leadership will act on the results of this survey? (1–5)

Uncomfortable to ask, and worth asking anyway. It measures the trust that every other answer depends on, and if you follow through it's the score that should rise between surveys.

10–11. Two open-ended questions

What is one thing we could do to make this a better place to work?

What do you enjoy most about working here?

The first gives you your action list. The second tells you what to protect when you change things. "One thing" matters: it pushes people toward their most important point instead of an exhaustive list.

A few rules for writing your own

• One idea per question. "How satisfied are you with your pay and benefits?" is two questions. Someone who loves the benefits and resents the pay can't answer it.
• Keep scales consistent. If most questions use 1–5 where 5 is best, don't flip one to 5-is-worst. People skim, and mixed scales produce answers that mean the opposite of what the respondent intended.
• Don't lead. "How much do you appreciate our new flexible schedule?" assumes the answer. "How has the new schedule affected your work?" doesn't.
• Skip what you won't act on. Every question costs the respondent time and costs you trust if it disappears without a trace.

Make it anonymous, and mean it

Employees are being asked to critique the people who decide their raises. Most won't do that candidly unless they're confident the answers can't be traced back to them.

Not asking for a name isn't enough. In a small company, a few innocent-looking questions can identify someone: "Engineering, 5+ years, remote" might describe exactly one person. Some practical rules:

• Use broad groups only. Ask for tenure as a range (less than 1 year, 1–3, 4–7, 8+), not exact years. Only ask for department if you'll actually use it.
• Set a minimum group size. Don't report results for any group with fewer than about five responses. Merge small teams into a broader category instead.
• Handle open-text carefully. Comments can identify people through details: a project, an incident, a date. Summarize themes when you share, rather than circulating raw comments.
• Say plainly what you collect. For example: "This survey doesn't ask for your name or email. Results are reviewed in aggregate, and no team smaller than five is reported separately."

We go deeper on this in How to Run Anonymous Surveys Without Losing Trust or Useful Data.

Running it

Have a leader introduce it. A short note from the owner or head of the team, saying why you're asking, what happens next, and when people will hear back, does more for response rates than any reminder email.

Give it a window. One to two weeks is plenty. Send one reminder about halfway through, to everyone. With an anonymous survey you can't tell who hasn't responded, and you shouldn't try.

Make it easy to answer anywhere. Plenty of people will answer on their phone, between tasks or on a break. Check it on a phone before sending.

Pick a rhythm and keep it. A fuller survey once or twice a year, plus a short three-to-five-question pulse each quarter, works well for most teams. Ask the same core questions each time. The trend is far more useful than any single result.

Reading the results

Look at the spread, not just the average. An average of 3.5 can mean everyone is lukewarm, or that half the team is thriving and half is miserable. Those are very different problems. Look at the distribution before you look at the average.

Calculate eNPS the standard way. People who answer 9–10 are promoters, 7–8 are passives, and 0–6 are detractors. eNPS is the percentage of promoters minus the percentage of detractors, so it runs from −100 to +100. Passives count toward the total but neither add nor subtract.

Compare yourself to yourself. Industry benchmarks vary wildly with company size, sector, and how the question was asked. Your own trend is the number that matters. A score that rises from one survey to the next tells you more than how you stack up against a stranger's average.

Read every comment, then group them. Sort the open-ended answers into themes: workload, communication, pay, a specific process, tools. A theme mentioned by a third of respondents is a priority, even if the ratings looked fine.

Watch the "will leadership act" score. If it's low, people are telling you they've been here before. Following through on this survey is the fix, and the next survey will show whether it worked.

Close the loop

This step decides whether the next survey is worth running. Within two or three weeks of closing it, share a short summary with everyone who was asked:

• What you heard. The top few themes, good and bad, in plain language.
• What you're going to do. Pick one to three concrete actions, with an owner and rough timing. Fewer, finished changes beat a long list of intentions.
• What you're not doing, and why. If many people asked for something you can't do right now, say so. Silence reads as "we didn't listen"; an honest "not this year, here's why" reads as respect.
• When you'll check in again. Then do it.

Common mistakes

• Running it once. A single survey is a snapshot. The value comes from asking the same questions again and seeing what moved.
• Making it too long. Forty questions feel like homework, and the answers near the end get careless.
• Reporting tiny groups. Showing a breakdown for a three-person team exposes those three people, whatever the survey promised.
• Hunting for the author of a comment. If anyone tries to work out who wrote what, word gets around, and that's the end of honest feedback.
• Waiting months to share results. By then, people have assumed nothing happened.

Start from a template

If you'd rather not build it from scratch, SurveyZebra has a free employee satisfaction survey template with eight of the questions above, from eNPS through the two open-ended questions, set up as an anonymous survey. Add the clarity, tools, or "will leadership act" questions if you want them, rename anything that doesn't fit your team, and send the link.

SurveyZebra doesn't store IP addresses, device details, or other tracking data with responses, so the survey is anonymous unless you add a question that asks who someone is. The results dashboard and CSV export are free, with up to 100 responses per survey. Larger teams can upgrade for unlimited responses.

Whatever tool you use, the survey is the easy part. What earns honest answers next time is what you do in the few weeks after it closes.