Back to blog
·15 min read

Employee engagement survey: the complete guide, from questions to action plan

A well-run employee engagement survey tells you what is working in your company, what is wearing the team down and where to act first. A poorly run one does the opposite: people answer with fear or indifference, the numbers say nothing and, since nothing happens afterwards, fewer people answer the next one.

This guide walks through the full process, start to finish: the goal, the dimensions, the questions, anonymity, timing, communication, analysis, presenting results and the action plan. It works for a 20-person company and for a 2,000-person one. At the end there is an eight-week calendar and a checklist to review before sending.

I write from organizational psychology and from HR, where I saw engagement surveys that changed things and others that only filled a slide deck. The difference was almost never the questionnaire: it was what happened before and after.

What workplace climate is and what an engagement survey measures

Workplace climate is how people perceive their work environment: how their manager treats them, whether they know what is going on, whether their effort is valued, whether they can grow, whether their pay feels fair, whether they can count on their coworkers. It is not the objective reality of the company, it is the perception of the people who work there. And that perception decides whether someone stays, recommends the company or does the bare minimum.

Climate and engagement are often used as synonyms, but they are not quite the same: climate measures the environment (the conditions), and engagement measures the person’s response to it (their willingness to go the extra mile, stay and recommend). A good survey usually measures both: several climate dimensions plus two or three engagement questions, such as eNPS.

What it is for

  • Finding problems before they become resignations. Voluntary turnover often shows up in the engagement survey a year earlier.
  • Learning which teams or managers have a different climate. The company average hides big differences.
  • Deciding where to invest: training, communication, pay, leadership.
  • Measuring whether what you did last year worked.
  • Giving people a channel to say what they would not say in a meeting.

If the company is not willing to act on the results, hold off on the survey. Asking and not acting is worse than not asking: it teaches the team that their opinion changes nothing.

Before you start: the goal and leadership commitment

Before writing a single question, answer three things in writing and agree on them with leadership:

  1. Why we are running the survey. "Understand the climate" is too vague. Better: "understand why turnover went up in operations" or "set a baseline before changing the way we work".
  2. What we are willing to do with the results. You do not need to know the actions yet, but there must be time and budget for at least one or two.
  3. Who sees what. Whether managers will see their team’s results, from how many responses, and whether anyone will see unfiltered comments. This is decided before, not when the data comes in.

Leadership commitment is not a formality. If the CEO introduces the survey and later presents the results, participation rises and actions get done. If the survey "belongs to HR", it becomes one more chore.

Dimensions: what to measure

A dimension (or factor) is a group of questions that measure the same topic. Grouping helps in two ways: results are much easier to read (ten dimensions instead of 40 separate questions) and each topic is measured with more than one question, which makes it more reliable.

There is no official list, but these dimensions show up in almost every engagement model:

  • Leadership or direct manager: feedback, respect, recognition, approachability.
  • Communication: hearing about changes in time, being able to speak up, information flowing between teams.
  • Recognition: whether effort is valued and achievements are noticed.
  • Pay and benefits: whether pay feels fair and whether people understand how it is set.
  • Growth: opportunities to grow, training, career path.
  • Teamwork and collaboration: within the team and across teams.
  • Working conditions and tools: space, equipment, systems, workload.
  • Work-life balance: hours, flexibility, workload.
  • Pride and belonging: whether they would recommend the company, whether they feel part of it.
  • Trust in leadership: whether they believe in the direction and decisions of senior management.

Choose 6 to 10 dimensions based on your goal. If the survey was triggered by a specific problem (for example, turnover in one team), make sure the related dimension has enough questions.

How many questions

3 to 5 questions per dimension and 25 to 45 closed questions in total, plus 2 or 3 open ones. That is about 10 to 15 minutes. Any longer and people start giving the same answer to finish quickly (it shows in the analysis: whole rows with the same answer). A dimension with a single question is not reliable; with more than six, you are probably asking the same thing in other words.

If you already have separate questions and want to organize them:

Engagement survey factors: how to group questions into dimensions

How to write the questions

Engagement surveys usually use statements with a 5-point agreement scale: Strongly disagree, Disagree, Neither agree nor disagree, Agree, Strongly agree. It is easy to answer, easy to analyze and lets you compare dimensions.

Rules for every statement

  1. One idea per question. "My manager gives me feedback and recognizes my work" is two questions: someone can get feedback and never any recognition.
  2. First person, about their own experience. "I get the training I need" instead of "The company trains its employees well". People know what they experience, not what others do.
  3. Positive wording. Negative statements ("I don’t feel heard") are confusing on an agreement scale: disagreeing with something negative is a double negative.
  4. Concrete. "I hear about changes that affect my work in time" can be answered; "Communication is good" gets answered with the mood of the day.
  5. No loaded words or absolutes. No "excellent", "always" or "never" inside the question.
  6. Words the whole team uses, without internal acronyms that frontline staff or new hires do not know.

Before closing the questionnaire, run it through the question checker: it flags double-barreled, leading and double-negative questions and unbalanced scales. These mistakes slip in even for experienced people.

Paste your questions and check the most common mistakes:

Survey question checker (free, no AI)

5 or 7 points

For engagement surveys, 5 points is enough and the most common, which makes comparisons easier. What matters is keeping the same scale on every question and every survey: if you change scales from one year to the next, you can no longer compare.

The full comparison:

Likert scale: 5 or 7 points?

Engagement questions

Add two or three questions that summarize engagement. The most common is eNPS: "On a scale of 0 to 10, how likely are you to recommend this company as a place to work to a friend or family member?". Others: "I see myself working here two years from now" and "I am proud to work here". They act as the outcome: later you can see which dimensions are most related to them.

To calculate eNPS from your responses:

eNPS calculator

Open questions

Two or three, at the end: "What do you value most about working here?", "What one thing would you change to work better?" and, if you like, "Anything else you want to tell us?". They explain the numbers. More than three gets tiring and answers get shorter.

Ready-to-copy questions by dimension:

Employee engagement survey question bank

Demographics and anonymity

To analyze by group you need a few demographic questions: department, location, tenure, job type. Each one makes the analysis richer, but also makes it easier to identify someone. If Finance has only one woman with more than ten years at the company, three data points identify her.

  • Only ask for the data you will use in the analysis.
  • Use ranges instead of exact values: tenure "less than 1 year, 1 to 3, 4 to 10, more than 10", not the start date.
  • Merge small teams. If a team has 3 people, combine it with a similar one for the survey.
  • Never show results for a group with fewer than 5 responses. It is the single most important anonymity rule.
  • Review open comments before sharing them: sometimes they include details that reveal who wrote them.

Anonymous or confidential

An anonymous survey cannot be linked to a person: everyone gets the same link and no name or email is asked. A confidential one does know who answered, but commits to not showing it. Both are fine. What is not fine is saying "anonymous" when it is confidential. If people find out, they will not trust any survey again.

A detail few people watch: if the link goes out in personalized emails with tracking, the tool knows who answered. For a truly anonymous survey, share one general link (by email, internal chat or a QR code on the floor).

And the ethical principles behind it:

Survey ethics: 8 rules

When and how often

Most companies run a full survey once a year, always at the same time so results can be compared. In between, short pulse surveys (5 to 10 questions) every three months help track the action plan topics.

  • Avoid year-end close, inventory or peak season: people answer in a rush or not at all.
  • Avoid launching right after big news (layoffs, a new manager, raises). You measure the reaction to that event, not the climate.
  • Do not tie it to performance reviews: people mix the two.
  • Keep the survey open for two weeks. With one, rotating shifts and vacations get left out; with more than three, momentum fades.

For follow-up between one full survey and the next:

Pulse surveys for teams

How to announce it and raise participation

Participation is the survey’s first result. Below 50%, results mostly represent the people who wanted to speak (often the happiest or the most upset). A realistic goal is 70% to 85%.

  1. Announce the survey a week ahead, from leadership: what it is for, how long it takes, that it is anonymous and what happened with the last one.
  2. Let people answer during working hours. If it has to be done at home, the message is that it does not matter.
  3. Offer more than one channel: link by email, internal chat and a QR code in common areas for people without a computer.
  4. Share participation progress by team (no names) halfway through. It works better than generic reminders.
  5. Send two reminders at most, with different messages.
  6. Do not pressure anyone or tie the survey to big prizes: you get quick, insincere answers.

To calculate it: complete responses / people invited × 100. Calculate it by team too: a team with 30% participation needs a more careful reading than one with 90%.

To calculate it by channel or team:

Response rate calculator

How to analyze the results

1. Check the data before calculating

Remove test responses, incomplete ones that did not reach halfway and those that picked the same answer on every question in very little time. Note how many you removed and why.

2. Calculate percent favorable for each question

Percent favorable is the share of people who answered Agree or Strongly agree. It is easier to understand than an average of 3.7 and shows differences better. Calculate percent unfavorable too (Disagree and Strongly disagree): a question with 60% favorable and 35% unfavorable is very different from one with 60% favorable and 5% unfavorable.

3. Group by dimension and build the index

Each dimension’s index is the average of its questions. The overall index is the average of the dimensions. As a common reference, 75% favorable or more is a strength, 60% to 74% is one to watch and below 60% is a priority. It is not a standard: the most useful comparison is with your own previous survey.

Paste your results and get the index, strengths and priorities:

Employee engagement index calculator by dimension

4. Compare by team, location and tenure

This is where the important findings appear. A company average of 70% can hide a team at 45%. Only show groups with 5 or more responses. And be careful not to read every difference as real: with small groups, an 8-point gap can be chance.

How to tell a real difference from noise:

Differences between teams: when they are real

To test a specific difference:

Compare two percentages

5. Compare with the previous survey

You can only compare questions whose wording and scale did not change. Look for changes of 5 points or more; 1 or 2 points is usually noise.

6. Read the open answers

Group them by theme (for example: pay, workload, manager, communication) and count how many fall into each. That way you can say "38 of 120 comments mention workload" instead of picking the most striking quotes. Quotes illustrate; they do not conclude.

7. Look for what drives engagement

If you included eNPS or an engagement question, check which dimensions relate most to it. A low dimension that is also strongly related to engagement is the first candidate for the action plan.

The step-by-step analysis, for any survey:

How to analyze survey results: the complete guide

How to present the results

Results are presented at three levels, in this order and within a few weeks:

  1. To leadership: the overall index, the dimensions, differences between teams and the three priorities you propose. A one-page summary before the detail.
  2. To managers: their team’s results compared with the company, with a guide on how to discuss them with their team. Prepare them: nobody likes receiving a 45% without context.
  3. To the whole team: what came out (good and bad, without spin) and what will be done. This is the part most often skipped and the one that most affects next year’s participation.

Start with what is going well. Not to sugarcoat, but because strengths also need protecting, and because people listen better to priorities after seeing their positive answers were taken into account too.

The action plan

The action plan turns the survey into something the team can see. Three rules make it work:

  1. Few priorities. One to three per cycle. Pick the lowest dimensions, the ones that dropped most or the ones most related to engagement.
  2. Concrete actions with an owner and a date. "Improve communication" is not an action. "A 15-minute meeting every Monday per team to share the week’s changes, owned by each team lead, from December 1" is.
  3. An indicator. Which question in the next survey should improve and by how much, or what you will count in the meantime (meetings held, people trained).

Build the plan with a small group from the team, not only with leadership. They know better than anyone why a question scored low, and the actions they propose get done more often.

Write your actions and download the plan with tracking:

Action plan generator (Excel and Word, free)

Example actions by dimension:

What to do with employee survey results

Common mistakes

  • Running the survey without agreeing first on what will be done with the results.
  • Promising anonymity and then showing results for groups of 2 or 3 people.
  • An 80-question survey because "while we are at it, let’s ask everything".
  • Changing the questions every year and losing the ability to compare.
  • Looking only at the company average and not at differences between teams.
  • Reading every 3-point difference as real.
  • Presenting results only to leadership and never to the team.
  • An action plan with 15 actions that nobody finishes.
  • Trying to find out who wrote a comment. If it happens once, word gets around, and the survey loses all its value.

An 8-week calendar

  1. Week 1: goal, leadership commitment and rules on who sees what.
  2. Week 2: dimensions, questions and demographics. Questionnaire review and a test with 3 to 5 people.
  3. Week 3: announcement from leadership and channel setup (email, chat, QR).
  4. Weeks 4 and 5: survey open, with a participation update halfway and two reminders.
  5. Week 6: data cleaning and analysis by dimension, by team and of open comments.
  6. Week 7: presentation to leadership and managers. Choice of priorities.
  7. Week 8: presentation to the whole team with the action plan, owners and dates.

After that, review the plan’s progress once a month and run a pulse survey three or four months later with the questions from the priority dimensions.

Checklist before sending

  1. The goal is written down and leadership committed to act.
  2. 6 to 10 dimensions, with 3 to 5 questions each.
  3. Each question has a single idea, positive wording and first person.
  4. The same scale on every closed question.
  5. One engagement question (eNPS) and 2 or 3 open questions.
  6. Only the demographics you will use, in ranges.
  7. Minimum of 5 responses per group, agreed and explained in the introduction.
  8. The introduction says what it is for, how long it takes and how anonymity is protected.
  9. You tested the survey with 3 to 5 people and fixed what they did not understand.
  10. The date to present results to the team is already set.

If you would rather start with something ready, download the employee engagement survey template in Word (to print) or Excel (to tabulate), or use the online template, which calculates results by dimension and by team as responses come in.

Free, with questions grouped by dimension:

Employee engagement survey template in Word and Excel

And to compare engagement with turnover:

Employee turnover rate calculator