Engagement scores dropped? How to find out why

When engagement scores drop, first check that the drop is real: compare response rates, who answered and which groups moved. Then find where it fell, read the comments from those groups, and ask the people there what changed. The survey shows you where to look. The reasons come from talking to the people in the groups that moved.

Most HR teams skip straight to an action plan for the whole company. That usually means fixing the wrong thing, in the wrong place, a few months too late. Here is the order we would work in.

Is the drop real, or did different people answer?

Before you look for causes, rule out the survey itself. A score can fall without anyone feeling worse.

• Look at the response rate. If 80% answered last year and 55% this year, you are comparing two different groups of people. Find out who stopped answering. A team that used to respond and went quiet is a finding in its own right.
• Look at who answered. New joiners, a site you acquired, a team that moved in after a restructure: any of them can shift the average while nothing changed for everyone else.
• Watch the small groups. In a team of eight, one person moving from agree to disagree changes the result by more than ten points. Treat small-team swings as a reason to ask, and wait before reading them as a trend.
• Check the questions and the timing. A reworded item or a new scale breaks the comparison. So does running the survey in the week of a pay announcement or at the peak of the busy season.

If the drop survives these checks, it is real enough to investigate.

Where did engagement drop? Segment the results

A company average hides almost everything. Break the results down by department, site, role, tenure and manager, and look at which items fell. In most organisations a handful of groups account for most of the drop, and the rest are flat.

Pay particular attention to teams that changed manager in the last year. Gallup’s research finds that managers account for 70% of the variance in team engagement, so a new manager, an absent one or an overloaded one is often where the story starts. Gallup’s latest State of the Global Workplace report puts global engagement at 20% in 2025 and attributes most of the recent fall to lower engagement among managers themselves. If your managers are stretched, their teams feel it.

Also look at which questions moved. A drop concentrated in workload and resourcing items points somewhere very different from a drop in “I know what is expected of me” or in recognition.

What do the survey comments say?

Read the free-text comments, but only from the groups that dropped. Comments from the whole company will drown the signal. Sort them into rough themes by hand: staffing, a specific change, a manager, pay, tools, communication.

Be honest about what comments can and can’t tell you. They come mostly from people who already speak up. In many teams the box is empty or holds one line. And you can’t ask a follow-up question: if someone writes “nobody listens”, you don’t know who isn’t listening, or about what. Treat what you find as a first guess to test.

How do you find the root cause of falling engagement scores?

Go and ask the people in the affected groups. You have a few ways to do it.

• Focus groups are quick and cheap, but people are careful in front of colleagues, especially if the problem involves someone in the room or their manager.
• Manager conversations work when the manager is trusted. If the manager is part of the problem, people won’t say so to them.
• Skip-level or HR one-to-ones go deeper, and they are slow. An HR business partner can manage perhaps a dozen in a week, so a depot of 150 drivers takes months.
• With AI-moderated interviews, everyone in the affected groups gets a one-on-one conversation at the same time, and the answers come back grouped into themes.

Whatever you choose, ask open questions about one thing at a time. Some that work well after a drop:

Questions to ask the teams where scores fell

1. What has changed in your work over the past six months?
2. What part of your job takes more energy than it used to?
3. What gets in the way of doing your job the way you would like to?
4. How do you hear about decisions that affect your work?
5. When did you last feel listened to here? What happened?
6. If you could fix one thing tomorrow, what would it be?

This is the work Spradley is built for. Each person gets a one-on-one interview by text or voice, on their own phone or computer, and the interviewer asks a follow-up question on each answer. The report groups what people said into themes with the quotes behind them, so you can compare the ward or depot that dropped with one that didn’t. Numbers and ratings are never shown for fewer than five people, and a privacy judge checks each written finding. More on the employee engagement page, and on how the two methods compare in AI-moderated interviews vs surveys.

What are the common causes of a drop in engagement scores?

Every organisation is different, but the same few causes come up again and again when you ask:

• Workload: vacancies not filled, people covering for colleagues, overtime that became normal. On a hospital ward this often shows up as staff being moved to cover other wards at short notice.
• A change nobody explained: a restructure, a new system, a new shift pattern. People put up with a lot of change. They struggle when nobody tells them why it happened or what comes next.
• A new manager. A well-liked manager leaves, or a new one arrives with a different style and no time to build trust.
• Pay and progression: a pay round that landed badly, or a promotion route that closed.
• Nothing happened after the last survey. Gallup warns that when leaders run a survey and take no action, engagement will likely decrease. Sometimes the drop is a verdict on the previous round.

What should you not do after engagement scores drop?

Don’t hunt for who said it

Trying to work out which person wrote a comment, or drilling down until a group is three people, destroys trust faster than any low score. People notice, and next year they answer with whatever is safe or don’t answer at all. Keep a minimum group size and stick to it. Our post on the employee survey anonymity threshold covers how to set one.

Don’t wait for next year’s survey

If you wait twelve months to see whether things improved, the team you are worried about will have changed, and some of the people in it will have left. Ask the affected groups now, act on one or two things, and check again in that group within a few months.

Don’t launch a company-wide plan before you know the cause

A wellbeing programme for everyone won’t fix a roster problem at one depot. Act where the drop is, on the cause people named.

What happens after you find the reasons?

Tell the people you asked what you heard, in their words, within a few weeks. Say what you will change, what you won’t, and why. Then ask the same group again after the change has had time to land. A short pulse or a second round of interviews will tell you whether it worked. If you are rethinking the yearly cycle altogether, see our guide to alternatives to the annual engagement survey. If your headline number is eNPS, read what eNPS does and doesn’t tell you.

Frequently asked questions

Key terms

response rate

The share of people invited who answered. A large change in response rate can move a score without anyone feeling differently.

segmentation

Breaking survey results down by group, such as department, site, role, tenure or manager, to see where a score moved.

engagement survey

A survey that asks every employee the same fixed items on a scale, usually once or twice a year, so scores can be compared over time and between groups.

skip-level meeting

A one-to-one between an employee and their manager’s manager, or someone from HR, without the direct manager present.

minimum group size

The smallest number of responses a group needs before its results are shown. Five is a common default; smaller groups are hidden or combined with a larger one.

AI-moderated interview

A one-on-one conversation run by software, in which the interviewer asks open questions and a follow-up question based on each answer. Everyone invited can take part at the same time.