Employee Engagement Metrics: What to Measure, Formulas and One Dashboard

Ask ten companies how they measure engagement and you will hear ten versions of the same answer: “we run a survey once a year”. One survey gives one number, and one number is easy to misread. In my HR metrics course I teach a different habit: measure what people say and what people do, and look at both on the same page.

Below are the metrics I use, the formula for each, and the way to collect them into one dashboard. If you want to try it on your own numbers right away, go to the free engagement mini-dashboard.

Engagement metrics are numbers that show how much people want to work well and want to stay. Before you count anything, separate four words that are constantly mixed up:

• Motivation — the desire to work better and more.
• Loyalty — the desire to stay with the company.
• Satisfaction — a positive assessment of working conditions.
• Engagement — motivation plus loyalty.

They do not replace each other. A satisfied person can be perfectly comfortable and make no effort at all. A motivated person can work hard and still leave next month. That is why “engagement metrics” is a set: each number catches one part of the picture.

I split them into two groups.

Survey metrics — what people say:

1. Engagement score — the average score per question of an engagement questionnaire.
2. eNPS — loyalty to the company as an employer.
3. Employee Satisfaction Index (ESI).
4. Intention to stay.
5. Survey participation rate.

Behaviour metrics — what people do:

1. Absence rate.
2. Turnover rate.
3. Employee referrals — how many candidates your people recommend for open vacancies.

There are more indirect signals: falling performance, people leaving earlier than they used to, a change in what customers say in their own surveys. I treat all of them the same way: a behaviour metric tells you that something may be wrong. You see the number, then you go and collect opinions.

Engagement score. Take a tested engagement questionnaire — the best known has 12 statements and a five-point scale. For every statement calculate the average answer, for each department and for the company:

The company-wide average is the least interesting line. The value is in the ranking: sort the questions from the lowest score to the highest. The questions at the bottom are your priority, and the wording of the question usually tells you what to do. If people give a low score to “someone encourages my development”, such a person has to appear. The questionnaire itself and the launch are covered in the employee engagement survey template, so I will not repeat them here.

eNPS. One question — how likely are you to recommend the company as an employer — on a scale from 0 to 10. Those who answer 9–10 are promoters, 7–8 are neutral, 0–6 are detractors.

The result runs from −100 to +100. If 124 people answered, 52 of them are promoters and 31 are detractors, eNPS = (52 − 31) ÷ 124 × 100 = +17. How to word the question and what counts as a good score is in the separate guide to the Employee Net Promoter Score.

Employee Satisfaction Index. Three questions, each on a ten-point scale: how satisfied are you with your workplace, how well does it meet your expectations, how close is it to ideal.

Example: the average answers are 7.4, 6.8 and 6.2. Their mean is 6.8, so ESI = (6.8 − 1) ÷ 9 × 100 = 64.4. The formula assumes the lowest possible answer is 1; then the index runs from 0 to 100.

Intention to stay. One statement with five answers from “strongly agree” to “strongly disagree”: “If everything depended only on me, I would work in this company for at least one year from now.” I use it because it helps to predict turnover before the resignations appear. The metric is the share of people who agree or strongly agree.

If 200 people were invited and 124 answered, participation is 62%.

Not everyone will answer, and managers will use that against you: “if not everybody replied, how can we trust it?” You do not need every answer. You need a representative sample: from a certain point, extra answers barely change the picture. Online sample-size calculators give the minimum. For a company of 1,000 people, at the usual settings of 95% confidence and a 5% margin of error, the minimum is 278 answers.

Two practical points:

• The sample works for the company, not for each team. 278 answers describe the thousand. They do not describe a department of 20 where four people replied.
• Participation is produced by reminders. The rule of thumb I show in the engagement course: without reminders expect about 10%, and every reminder adds roughly another 10 points — about 50% after four.

Participation is also a signal in its own right: a team that used to answer and has stopped is worth a conversation before you look at any score.

Absence rate shows what share of working time people were not at work.

Count illness, time off and no-shows. Do not count vacation: it is legitimate absence, and for the company it is rather a plus. Example: a team of 200 people has 21 working days in the month — 4,200 scheduled days. People were absent for 126 of them. Absence rate = 126 ÷ 4,200 = 3.0%.

Turnover rate. When turnover goes up, loyalty is going down — that is why turnover belongs on the engagement dashboard and not only in the retention report.

Three leavers in a month with an average headcount of 200 is 1.5% for the month, or 18% a year at the same pace. Monthly, voluntary and first-year versions, with a calculator, are in the guide to the employee turnover rate.

Employee referrals. The more candidates your people recommend for open vacancies, the higher their loyalty; when recommendations dry up, loyalty is falling. Count the number of referred candidates per quarter and watch the direction.

If your company measures customer loyalty or customer satisfaction, put that line on the same page. Employee loyalty is usually considered directly linked to customer loyalty, so a drop there is one more indirect signal.

A dashboard here is one page — a spreadsheet is enough. Rows are the metrics. Columns are the current period, the previous period and each department.

1. Choose three or four survey metrics and two or three behaviour metrics. The list above is a menu, not a requirement.
2. Write the formula next to each one. Who counts as invited, which absences are included, how average headcount is calculated. Anyone should be able to recalculate and get the same number.
3. Set the rhythm. A full survey once or twice a year shows the company as a whole. A pulse survey — a few questions in a form, an HR system or a chat bot — collects data continuously, so you can react at the level of a team and return to any period later. Behaviour metrics are counted monthly. My rule for survey length: one screen, once a month.
4. Segment. By department first, then by length of service and by manager or non-manager, as in the Wargaming example.
5. Cross the two groups. Split turnover by engagement and by loyalty: are the people who leave the ones who scored low?
6. Add a written conclusion. For every metric answer four questions: what do the figures say, how do they differ from the norm, how do we bring them back, what actions could be taken.

A short loyalty or pulse report fits on one or two pages. A report on a full engagement or satisfaction survey takes five to ten: purpose, audience, sample, method, findings with segmentation, recommendations, next steps.

These are the combinations I look for.

• High score, low participation. The score belongs to the people who answered. Find out who stayed silent before you celebrate.
• Good survey numbers, rising absence or turnover. Believe the behaviour and check the instrument.
• High satisfaction, low engagement score. People are comfortable. That is not the same as wanting to do the job well.
• eNPS falling in one department while the company number is flat. The average hides it; the breakdown shows it.
• Everything stable, referrals falling. A quiet early signal about loyalty.

The second combination deserves a separate note. More than once I have met this situation: a company says, “we have an engagement questionnaire”. Great — where is it from? Who developed it, who tested it? Nobody knows, we wrote it ourselves. They take a random set of questions and announce that engagement is 90% — while turnover is high and people are unhappy. It is like measuring the temperature of a room with a ruler. You will get a number, you can even say “twenty-two”, but the instrument is not valid, because a ruler measures something else.

A metric by itself changes nothing. If you want engagement, satisfaction or loyalty to grow, you have to turn bad answers into good ones: take the questions with the lowest scores, read the comments people wrote, plan actions for those questions. When the worst answers improve, the overall picture improves.

The metrics show where to look; the work starts after that. Ready-made questionnaires, report examples and the rest of the HR numbers — recruitment, pay, development, retention — are in my HR Metrics Course: Analytics and Dashboards for HR. What to do with the results, team by team, is the subject of the Employee Engagement Course: Gallup Q12 and eNPS.