Skip to main content

Employee Engagement

How to Write Employee Survey Questions

Employee Engagement11 min read

Good employee survey questions ask about one thing at a time, describe observable experience rather than feelings in the abstract, and connect to a decision you are prepared to make. This guide covers how to write them, with 60+ examples organised by engagement driver.

Question quality determines data quality more than sample size, platform or timing. A poorly worded question produces a number you cannot act on.

The six rules of writing survey questions

1. One thing per question

Double-barrelled questions are the most common error and the hardest to spot in your own draft.

Double-barrelled (avoid)Single (use)
“My manager gives regular feedback and supports my development”“My manager gives me feedback I can act on”
“My manager supports my development”
“I have the tools and training to do my job”“I have the equipment I need”
“I have been trained for everything I’m asked to do”
“Communication is clear and timely”“I understand the information I receive”
“I find out about changes before they affect me”

The test: if someone could agree with half your question and disagree with the other half, split it.

2. Ask about experience, not opinion

“Do you think recognition is important?” measures nothing useful. “In the last month, has anyone told you they appreciated your work?” measures something you can act on.

3. Avoid leading language

“How helpful was the new rostering system?” presumes helpfulness. “How has the new rostering system affected your work?” does not.

4. Use consistent scales

Pick one agreement scale and use it throughout. Mixing agreement, frequency and satisfaction scales in one survey makes results harder to compare and the survey harder to complete.

A five-point agreement scale (strongly disagree to strongly agree) is the workhorse. Five points is enough granularity without the false precision of ten.

5. Write for the person answering

“I feel psychologically safe raising concerns with leadership” is a phrase from a research paper. “If something wasn’t right, I’d feel comfortable saying so” measures the same thing and makes sense to a kitchen hand on a break.

6. Only ask what you will act on

Every question implies a willingness to respond. If a low score on a question would change nothing, remove it — asking and then ignoring is worse than not asking.

Common mistake

Writing questions in HR language. “I am empowered to leverage my strengths in alignment with organisational objectives” will produce responses, but nobody will be answering the same question. Read every question out loud and ask whether you would say it to someone on shift.

60+ questions by engagement driver

Take three to five from each driver you want to measure. A comprehensive survey lands around 30–40 questions and under five minutes.

Manager effectiveness (8)

Role clarity (6)

Recognition (6)

Development and growth (7)

Communication (6)

Voice and action (5)

Prosper Employee Engagement Use these questions in a survey that reaches everyone Prosper delivers surveys by push notification, SMS or email — so the people who have never had a company email address answer too. Results segment by site, team, role and tenure. See Prosper Employee Engagement

Wellbeing and workload (6)

Team and belonging (5)

Frontline-specific (8)

These do not appear in standard question banks and are among the most useful for shift-based operations.

Open-text questions (5)

Two open questions produce more actionable information than twenty scaled ones. Include them, and read them.

Question types and when to use each

TypeBest forWatch out for
5-point agreementMost engagement driversCentral tendency — people cluster on 3
Frequency (never–always)Behaviour, e.g. recognitionRecall accuracy over long periods
0–10 (eNPS)Single headline metricTells you the temperature, not the cause
Yes / NoFactual checks, e.g. “have you had a 1:1?”No nuance — use sparingly
Open textUnderstanding whyAnalysis effort at scale — needs AI theming

For eNPS specifically, see our glossary entry on eNPS and the guide to engagement survey software.

How many questions should a survey have?

Survey typeQuestionsCompletion time
Comprehensive engagement30–404–5 minutes
Pulse5–10Under 2 minutes
eNPS only1–2Seconds
Onboarding / lifecycle8–122–3 minutes
Exit10–153–4 minutes
Expert tip

Time your own survey before sending it. Complete it on a phone, standing up, as if you had five minutes between tasks. If it feels long to you sitting at a desk, it is far too long for someone on a break.

Common mistakes

Writing new questions every cycle. Consistency is what makes trend data possible. Keep a stable core and vary a small number.

Too many questions. Fewer questions completed by 80% of people beat more questions completed by 40%.

Asking about things you cannot change. It raises expectations you will not meet.

No open-text questions. Scores tell you what; comments tell you why. Skipping them removes the diagnosis.

Using a template unchanged. Standard banks are written for office workforces. The frontline questions above will not appear in one.

See it in practice Survey questions that reach the whole workforce Ready-made question sets designed for frontline teams, delivered by push notification or SMS, with AI theming across every open-text response. Explore Prosper Employee Engagement

Frequently asked questions

Ask about one thing at a time, describe observable experience rather than abstract feelings, avoid leading language, use a consistent scale, write in plain language your workforce actually uses, and only ask what you are prepared to act on.

30 to 40 for a comprehensive survey, completable in under five minutes. Pulse surveys should be 5 to 10 questions and under two minutes.

One that asks about two things at once — for example "my manager gives regular feedback and supports my development". A middling answer tells you nothing about which half is the problem. If someone could agree with half and disagree with the other half, split it.

Yes. Two open questions usually produce more actionable information than twenty scaled ones. Scores tell you what is happening; comments tell you why.

A five-point agreement scale used consistently throughout. Five points gives enough granularity without the false precision of ten, and mixing scale types makes results harder to compare.

Questions grounded in the shift — whether they are told about changes before those changes affect them, whether handover works, whether they have what they need to start a shift, and whether head office understands the job. These rarely appear in standard question banks.

No. Keep a stable core set so results can be compared over time, and vary only a small number based on what you are currently working on.

Remove any assumption from the wording. "How helpful was the new system?" presumes helpfulness; "How has the new system affected your work?" does not.

See it in practice

Ask better questions, and reach everyone with them

Question sets designed for frontline teams, delivered by push notification or SMS, with AI theming across every open-text response.

Surveys that reach everyonePush and SMS delivery, no work email needed
See how it works