Employee survey questions: a practical research guide

Most employee surveys try to measure everything at once, which is exactly why the results are hard to act on. A tighter approach works better: pick two or three research goals for the current cycle — understanding turnover risk on one team, say, or checking whether a new manager-training program changed day-to-day behavior — then choose 15 to 25 questions that speak directly to those goals.

The question bank below is organized by what each question measures, not by department or job level, so you can pull individual items into whatever survey tool you already use rather than importing a fixed questionnaire wholesale.

• Treating “employee survey” as one undifferentiated topic is how surveys end up vague. These are the areas that tend to produce distinct, usable signal when kept separate:
• Role clarity — whether people understand what's expected of them day to day.
• Manager relationship — trust, feedback quality, and whether a manager removes obstacles or creates them.
• Growth and development — whether people can see a path forward, promotion or not.
• Recognition — whether good work gets noticed, not just mistakes.
• Workload and boundaries — whether the job is sustainable at its current pace.
• Communication — whether information reaches people before it reaches the rumor mill.
• Trust in leadership — whether people believe leadership tells the truth, including unwelcome truth.
• Overall sentiment — a single summary question, often phrased as likelihood to recommend the organization as a place to work.

• “I understand what is expected of me in my role.” (5-point agree/disagree scale)
• “In the last three months, I received feedback from my manager that helped me improve.” (agree/disagree)
• “My manager gives me credit for the work I do.”
• “When priorities change, I hear about it from my manager before I hear about it elsewhere.”
• “My manager asks for my input before making decisions that affect my work.”

• “I can see a path for growth here, even if it isn't a promotion.” (agree/disagree)
• “I've had a conversation about my career goals with my manager in the past six months.” (yes/no)
• “My workload is manageable most weeks.” (agree/disagree)
• “I feel recognized for good work, not just called out for mistakes.”

• “Leadership explains the reasoning behind major decisions, not just the decisions themselves.”
• “I know where to find accurate information about changes that affect my job.”
• “I trust that leadership tells employees the truth, even when the news isn't good.”
• “This organization does what it says it will do.”
• “If I raised a concern, I believe it would be taken seriously.”
• “On a scale of 0 to 10, how likely are you to recommend this organization as a place to work?” (the standard eNPS phrasing)

A 5-point agreement scale (strongly disagree to strongly agree) is easier for people to answer consistently than a 7- or 10-point scale, and simpler to report on afterward. Label every point instead of just the two ends — an unlabeled midpoint gets interpreted differently by different respondents.

Avoid double-barreled questions that ask about two things at once, such as “my manager communicates clearly and treats me fairly.” A respondent who agrees with one half and not the other has no accurate way to answer.

Only mix positively- and negatively-framed statements if your survey tool correctly reverse-scores the negative items during analysis. If it doesn't, keep every item framed in the same direction — inconsistent framing without correct reverse-scoring will quietly distort the results.

Somewhere past 20 to 25 questions, completion rates start dropping and the last third of the survey gets rushed, which shows up as flat, low-effort answers (“straight-lining”) rather than genuine responses. If a topic needs more than 25 questions to cover properly, it usually needs to be split into a separate, shorter survey instead.

Timing matters as much as length: sending a survey during a known busy period — close of quarter, a seasonal crunch, right after layoffs — will depress both response rate and scores for reasons that have nothing to do with the questions themselves.

One or two open-text questions are usually enough for a single survey. “What's one thing that would make your work easier?” tends to produce more usable answers than “do you have any other comments?” because it gives people a specific frame to respond within.

Open text takes longer to read and categorize than closed questions. If the goal is tracking a metric across multiple survey cycles, let closed questions carry that weight and use open text to catch what the scales missed.

If people believe their individual answers can be traced back to them, scores skew positive and open-text answers get shorter and safer. True anonymity is hard to guarantee on a team of four, no matter what the survey tool promises — anyone can work out who wrote a comment about “my manager” when there's only one manager and four possible respondents.

Where full anonymity isn't realistic, say so plainly in the survey instructions rather than promising something the sample size can't support. Confidential-but-not-anonymous — results seen only by HR, not by direct managers — is often the more honest option for small teams.

A survey with a low response rate, or one where entire teams opted out, can't be read as representative of the whole organization — treat it as a signal from the people who responded, not a census.

Watch for question-order effects: a satisfaction question placed right after several pointed criticism questions will often score lower than the same question asked first. If you're tracking a metric over time, keep its position in the survey consistent between cycles.

Small groups within a team, roughly under five respondents, usually need to be suppressed or folded into a broader category in reporting, so individual answers can't be worked out by elimination.