An employee survey is quick to send out. The hard part comes afterwards: when the response rate stays thin, the report sits in a drawer and hardly anyone bothers next time round.
You run an employee survey in five phases: set the objective and decide who takes part, write the questionnaire and commit to anonymity, invite people and keep an eye on the response rate, analyse the answers and report them back, then turn what you found into visible action. By the end you have a plan that holds up in a team of 20 as well as in a department of 500.
With empirio.ai, an online survey tool from Germany, EU hosting is already part of the free plan. That removes one of the points where an employee survey tends to stall early on. What happens to the answers afterwards is a decision no tool makes for you.
📌 The key points at a glance
- An employee survey runs in five phases, from planning to action.
- The HSE calls a survey the start of a risk assessment, not the whole of it.
- Consult staff representatives and trade unions before the questionnaire is final.
- Skip the feedback session and the next response rate falls.
- Repeat measures need the same statements and the same scale.
Create a survey for free
With empirio.ai you can create a modern online survey in minutes — with hosting in the EU.
- AI-built survey
- Adjust by drag & drop
- Real-time analysis
What an employee survey is good for, and what it is not
An employee survey gives you a picture of the workforce at one point in time, from everybody at once and in a form you can compare. It shows which themes are shared across the organisation and which ones only burn in one department. What a survey does not give you is the solution.
The Health and Safety Executive is blunt about that limit in its guidance on the Management Standards. A survey, it says, “is only the start of the risk assessment process and a broad indicator of the situation in an organisation”, and you “should not use surveys in isolation but with other sources of information”. The HSE Indicator Tool itself runs to 35 items about working conditions known to contribute to work-related stress.
That leads to a decision you make before the first question is written. Which kind of survey suits your reason for asking? Three formats have become standard and they solve different problems.
| Type of survey | When it fits | Usual rhythm |
|---|---|---|
| Full survey | A complete picture across every theme at work | every one to two years |
| Pulse survey | Tracking a trend, spotting early when something slips | monthly to quarterly |
| Topic survey | One specific trigger, such as a new hybrid working policy | one-off |
The three formats work together. Plenty of organisations run one large survey on a fixed annual cycle and slot short pulse checks in between, often around a single measure such as the Employee Net Promoter Score. The point is to know the difference before you plan: five pulse questions will never replace a full picture and a full survey makes a poor early warning system.
If what you actually want to know is how content people are with their work, their manager and their pay, the questions and scales for that sit in the article on how to measure employee satisfaction. This one stays with the process.
The five phases of an employee survey
An employee survey falls into five phases: plan, prepare, run, analyse and report back, then act. The order is not a formality. Each phase decides whether the next one can work at all, and the last one decides whether there is ever a second survey.
That the follow-up belongs to the instrument rather than being tacked on afterwards has been settled in the literature for years. The Handbuch Mitarbeiterbefragung edited by Michel E. Domsch and Désirée Ladwig (3rd edition, Springer Gabler 2013) treats feedback and action planning as project phases in their own right, each with its own owner. Plan only as far as the analysis and you have planned half the project.
| Phase | What happens in it | Who decides |
|---|---|---|
| 1. Plan | Objective, who takes part, timing, project team | Senior leadership |
| 2. Prepare | Questionnaire, pilot, anonymity promise, data protection | Project team |
| 3. Run | Invitation, fieldwork period, reminder, response rate | Project team |
| 4. Analyse | Report, feedback session with the workforce | Project team and managers |
| 5. Act | Actions, deadlines, checking whether they worked | Senior leadership and departments |
For a full survey, allow roughly three months from the first planning meeting to the feedback session. The questionnaire itself is the smallest item on that bill. The time goes on consultation, the pilot and the question of who gets to decide what afterwards.
Planning an employee survey: objective, participants and timing
Planning starts with a question that sounds as though it has already been answered: what will you do with the result? An objective along the lines of “we want to know how people are feeling” produces a questionnaire about everything and a report you cannot act on.
A workable objective names a decision. “We want to know whether the new shift pattern is holding up before we roll it out to the second site” is one of those. Who gets asked, what gets asked and when the result has to be on the table all follow from it almost by themselves.
Who takes part and who does not
The participant group normally covers the whole workforce, part-time staff, apprentices and people on parental leave included. Leave a group out and you get a skewed picture and a decision you then have to explain. Agency workers and contractors are the borderline case: they live the same working day but sit outside the payroll. Decide that deliberately and put the decision in the invitation.
Timing shapes the answers
A survey run in the week after a restructure measures the restructure, not how people feel about their work. The same applies to the days after a pay review, with the sign reversed. Put the survey into a quiet stretch of the year and keep roughly to that slot next time, otherwise you end up comparing two different moods instead of two measurements.
Consultation is not a formality in the UK
There is no statutory works council in Britain, so nobody has to sign the questionnaire off. That makes early consultation more important rather than less. The HSE advises organisations to “keep managers, staff representatives, trade unions and employees fully consulted and informed throughout the survey process so that the figures do not come as a shock”. A recognised union, an employee forum or a staff network belongs in the objective-setting conversation, not in the final sign-off.
One last item belongs in the planning, and it is the uncomfortable one. Agree who sees the results, when they see them and who decides on action, before the first answer arrives. Leave that open and the analysis lands in a question of ownership while the report sits around for three months.
Writing the questionnaire and promising anonymity
The questionnaire for an employee survey holds a short welcome text, several blocks of statements people agree or disagree with, a small number of background questions and one open question at the end. Each block belongs to one theme and each statement describes exactly one thing a person can observe.
Length is where most drafts fall over. Every extra question costs you responses, and not evenly: drop-offs cluster the moment people notice the progress bar barely moving. Cut any question you cannot name a consequence for. How to order the blocks sits in the article on survey structure, and the wording of the questions themselves in the guide to creating a questionnaire.
Pick one scale and stay with it
For statements about working life, an agreement scale is the standard choice, running from “strongly agree” to “strongly disagree”. Decide on a number of points and keep it within a block of statements. Mixed scales make the analysis harder than it needs to be and make year-on-year comparison impossible. How to analyse a scale like that sits in the article on the Likert scale.
A pilot costs two days and saves the survey
Give the draft to five or ten people from different parts of the organisation and ask them to think aloud as they go. Two things come out of that which you would otherwise miss: which wording gets misread and how long completion actually takes. Put that timing into the invitation afterwards, because an honest estimate lowers the barrier to starting at all.
Making anonymity hold up technically
A survey is only anonymous once nothing is stored in the background that ties an answer back to a person. Three settings decide that, and all three are fixed before you send: one participation link for everyone instead of personalised invitations, no mandatory login through the company account, and no IP address or device ID stored against an individual answer.
Those same three points are how you assess a survey tool, rather than taking the word “anonymous” in the product description at face value. Can participant tracking actually be switched off? Are the servers inside the EU? And is there a data processing agreement without you having to move up to an enterprise plan? If you cannot get a clear answer to those three, you cannot promise anonymity either.
With empirio.ai you share an employee survey as a link, and the data processing agreement is part of the free plan.
Careful
The promise of anonymity is where a survey can lose trust for good. Only promise what the survey tool actually delivers: no personalised links, no mandatory login, no IP logging. Agree in advance how small a group can be before its results stop being reported. How an anonymous survey is set up technically has its own article, and the UK GDPR side of it sits in the piece on data protection in surveys.
Create a survey for free
With empirio.ai you can create a modern online survey in minutes — with hosting in the EU.
- AI-built survey
- Adjust by drag & drop
- Real-time analysis
Which themes and questions belong in the questionnaire
An annual employee survey usually covers six themes: workload and the job itself, management and feedback, teamwork, information and communication, tools and processes, and development and prospects. Five or six statements per theme is enough; anything more you pay for in responses.
The themes are not a fixed syllabus. Which ones you take follows the objective you set in the planning phase: if you want to know whether the new shift pattern holds, workload and processes need to be thorough and development can wait until next time.
| Theme | Example statement | When it belongs in the questionnaire |
|---|---|---|
| Workload and the job | “My workload is manageable in a normal week.” | always |
| Management and feedback | “I get regular feedback on my work.” | always |
| Teamwork | “In my team, disagreements get raised openly.” | always |
| Information and communication | “I hear about decisions that affect me early enough.” | after restructures and changes in leadership |
| Tools and processes | “The way things are organised here helps me do my job.” | when process improvement is the objective |
| Development and prospects | “I can see a future for myself in this organisation.” | where turnover is high |
Ten questions on process improvement
If what you are after is how the work flows rather than how people feel, you ask differently. These ten lines cover the tools and processes theme and can be used almost unchanged. The first eight run on the same agreement scale; the last two are open questions.
- The processes in my area are clearly documented.
- I know who decides what on the tasks I work on.
- I rarely have to enter or pass on the same information twice.
- The systems I work with support the job I have to do.
- Queries to other departments get answered promptly.
- I have enough time to do my work properly.
- When a process does not work, I know where to report it.
- Suggestions for improvement from my area get picked up.
- Which step in your work regularly costs you the most time?
- What would you change first if it were up to you?
Those last two lines are the most valuable in the theme, because they supply the reason behind the numbers. Once the statements are settled you can enter them straight into the tool and make an online survey before the draft goes round for sign-off.
How to spot a question you can cut
Every statement in the questionnaire needs a consequence attached to it. Go through the list once before the pilot and ask yourself, line by line: what would we do if agreement here came out at 40 per cent? If no answer comes to mind, the question goes. It is an uncomfortable test, because every question has a sponsor somewhere in the building, and it is the only filter that reliably shortens anything.
Running an employee survey: invitation, fieldwork and response rate
The invitation to an employee survey names four things: why you are asking, how many minutes it takes, the anonymity promise and the date on which the results will be presented. The fourth is missing from most invitations and is the one that does the most work, because it turns the survey into an appointment.
A sample invitation you can adapt
Keep it short and send it from the person who will be standing in the room afterwards. A circular from HR carries less weight than four sentences from the person running the organisation.
Hello everyone,
we want to know how working together here is going at the moment and where it is not. It takes about eight minutes.
The survey is anonymous. We see no names and no email addresses, and we will not report results for any group of fewer than five answers. You can take part until [date].
We will present the results to everyone on [date], including the uncomfortable ones. What we do about them we will discuss afterwards in each department.
Thank you for taking part.
Three details in that template genuinely need checking before it goes out: the number of minutes from the pilot, the group size below which you stop reporting, and the date of the feedback session. Leave any of them approximate and it will be noticed, if not now then afterwards.
The response rate you need
Two weeks of fieldwork has proved a sensible length. Anything shorter misses part-time staff, shift workers and people coming back from leave; much longer adds little, because anyone who has not taken part after three weeks usually never will. One reminder around the halfway mark is generally enough.
No British regulator sets a binding minimum response rate for an employee survey. In practice the rule is simple: the further the response falls below half the workforce, the more open the question of whether the answers describe the organisation or only the notably content and the notably annoyed. Both groups answer more readily than everyone else. Record the response rate in the report, broken down by department.
A thin response is rarely a technical problem. Most of the time it is the answer to the last survey, after which nothing happened. That is almost impossible to repair in the current wave and entirely possible to repair in the next one.
Reminding people without leaning on them
Send the reminder to everybody, not only to those who have not answered yet. A targeted reminder means somebody is tracking participation, and that cuts against the anonymity promise even where the tool stores nothing but a completion flag. Word it as a note about the time left, not as a chase.
Check before you send
- The invitation names purpose, length, anonymity and the feedback date.
- Staff representatives have seen the questionnaire.
- The questionnaire works properly on a phone.
- No question collects a detail you will not analyse.
- The feedback session is in every manager’s calendar.
Analysing the results and reporting them back
The analysis of an employee survey answers two questions: where does the organisation stand compared with the last measurement, and which department sits furthest from the average. Everything beyond that is detail and can wait.
Keep the first report short for that reason. Three figures per theme are enough: the score, the spread of answers and the number of people who replied. Getting from raw data to that point is covered step by step in the article on how to analyse a questionnaire.
The free-text answers are where the value sits
Open answers cost reading time and give you the one thing no scale can: the reason. Read them all before you sort them and build the categories afterwards. Three or four themes usually keep coming back, and those are the ones that belong in the feedback, quoted as written and without smoothing.
The feedback session belongs in the first few weeks
Report the results within a few weeks of the survey closing, even when no action has been agreed yet. People read silence as an answer, and an unfriendly one. Show the uncomfortable numbers first: a presentation that opens with the good scores loses the room immediately.
There is no single right way to split the sessions. What works is a short all-staff walkthrough followed by separate sessions per department, where the local numbers get discussed. Group small units together at that point, even if it leaves one detailed question unanswered.
Turning results into action
Action comes out of an employee survey only when three things are settled: what will be done, who does it and by when. Miss one of those and what is left is an intention, and intentions rarely survive the next quarter end.
The temptation is to take on every weak spot at once. Take two or three themes that keep coming back in the free-text answers instead and work through them visibly. One solved theme does more for the next response rate than ten announced ones.
A survey is a promise. Asking the question says you want to hear the answer.
What checking the effect actually needs
Measure the effect in the next survey using the same statements, the same scale and a similar point in the year. Only then is the change readable. Rewrite the questionnaire between waves because you have since thought of better wording and you trade the comparison for the polish, which is a bad exchange.
“No” counts as feedback too
Not every request that comes out of a survey can be granted. An extra role, a move, a pay rise: some of it fails on budget alone. Say so openly, with the reason and without holding out hope. A reasoned no does far less damage to trust than a “we’re looking into it” that nobody ever hears about again.
Common mistakes in employee surveys
The four mistakes below cost nothing in technology and a great deal in trust, and three of them only show up at the next survey. They happen regardless of size, in a charity with twelve volunteers as readily as in a listed company.
What they share is the cause: the survey gets planned as data collection rather than as the opening of a conversation. Sending the link also promises that somebody will read the answers and do something with them.
The survey stands in for the conversation
A questionnaire is no substitute for management. Where a conflict has been simmering in a team for months, an anonymous survey does not resolve it, it merely puts it on record. Use the survey for what it does well, which is the view across a lot of people at once. Individual cases get sorted in conversation, whatever the survey calendar says.
The questionnaire grows during consultation
Every department involved wants one more question added and the draft ends up twice its original length. Set a ceiling before the consultation round and treat it as fixed. Anyone who wants a question in has to take one out. It sounds bureaucratic and it is the only rule that holds in practice.
Results only travel upwards
A report that goes to the board and nowhere else turns the survey into a monitoring exercise. The workforce notices at the latest when six weeks have passed in silence. Put the all-staff feedback in the plan as a fixed date, not as an optional extra once the analysis is done.
The next wave starts without a look back
Open the invitation to the second survey with what came of the first. Two sentences will do: this is what we heard, this is what we changed, this is what we did not change and why. Without that look back, people answer the same questions into a void a second time and the response rate will tell you so.
Conclusion
Plan the feedback before you write the questionnaire. An employee survey with 20 statements, two weeks of fieldwork and a fixed date for the feedback session will do more for you than any sophisticated instrument without a follow-up process. And measure the next one with the same statements, or you have two snapshots instead of a trend.
Where to go next
- Want to know which questions measure satisfaction? Measuring employee satisfaction
- Looking for one measure you can track wave by wave? Employee Net Promoter Score
- Need a ready-made starting point? Survey and questionnaire templates
Ready to get the survey out this week?
Set your employee survey up in empirio.ai, share it as a link and get the analysis per theme without building a single spreadsheet. Sign up free and get started.