Identify disengaged employees

Every engagement program eventually arrives at the same question: which of our people are quietly slipping away, and can we do something before they go? Disengagement rarely announces itself. It shows up as someone who stops answering surveys, a rating that drifts low, or an eNPS score that lands in detractor territory. Catching those signals early is one of the most valuable things an engagement platform can do for you.

This guide explains exactly how CultureMonkey surfaces at-risk employees, what each signal means, and - most importantly - the single boundary that governs the whole feature: anonymity. Because there is a hard rule at the center of this that you must understand before you go looking for names. On an anonymous survey, you cannot see who gave a low score, and no setting changes that. Named insight into individuals exists only for surveys that were deliberately run as non-anonymous. The rest of this article is about knowing which situation you're in and using each one honestly.

The three signals of disengagement

Disengagement is a pattern, not a single data point, but CultureMonkey watches for three concrete signals that reliably point at risk. Understanding each one, and what it does and doesn't tell you, is the foundation for everything else.

Each of these is a genuine early warning. A person who has gone silent across several surveys may be checking out. A cluster of low rating scores points at a driver that has broken down. A wall of detractors is a flight-risk signal you ignore at your peril. But the way you're allowed to act on each one depends on whether the survey that produced it was anonymous or not. That's the boundary we turn to next, because it changes everything.

The anonymity boundary: read this first

Here is the rule that shapes this entire feature. On an anonymous survey, individual answers are never joined back to a person - not for admins, not for anyone. You can see that a group of people scored low, but you cannot see which people. This is not a limitation you can configure away; it's enforced in the data layer, and it's the reason honest feedback exists at all. For the full picture of how this protection works, see How anonymity keeps your feedback safe.

So the same three signals behave in two completely different ways depending on the survey mode:

The distinction is deliberate. A confidential survey is one you chose to run with identity attached, precisely so you could follow up with individuals. Onboarding check-ins, lifecycle stage surveys, and manager one-on-ones are the classic cases: you want to reach out to a specific new hire who's struggling, and everyone understood that going in. An anonymous survey made the opposite promise, and CultureMonkey keeps it.

What you can see on an anonymous survey

Most engagement and pulse surveys are anonymous, so this is the situation you'll be in most often. You can absolutely identify where disengagement lives, just not who it is. And in practice, that's usually enough to act well.

Here's what's available to you:

• Low driver scores. The engagement drivers that score lowest tell you which parts of the experience are failing. A low "Recognition" or "Manager support" score is a disengagement signal at the group level.
• The eNPS detractor split. Your eNPS breakdown shows how many people are detractors (and the shape of the 0-10 curve), so you can see the size of the at-risk group without seeing its members.
• Low-scoring segments. Filtering scores by team, location, or manager shows which segments are struggling, as long as each segment meets the anonymity threshold.
• The comments behind the numbers. Open-text feedback, kept anonymous, is where the "why" lives. Detractor comments usually name the concrete problem.
• Aggregate participation. You can see how many people haven't responded and nudge exactly those non-responders, all without a name. See Identify and follow up with non-responders.

app.culturemonkey.io/reports

The move here is to treat the pattern as the target. A team with a low engagement score and a run of detractor comments about workload is a clear, actionable finding. You take it to that team's manager as a theme, agree on an action plan, and close the loop with the group. You never need a name to do that, and trying to get one only puts the honesty of future surveys at risk.

Named at-risk insight: how it actually works

For surveys you deliberately ran as confidential, CultureMonkey does offer per-employee at-risk lists. This is the "named insight" feature, and it's important to be precise about what it is, where the data comes from, and what gates it.

The feature surfaces the same three categories, now as lists of actual people:

• Silent employees - "Employees who have not answered [this] survey." The list is built from participant records that were sent the survey but have no submission.
• Low engaged employees - "Employees who have provided low scores for rating questions," meaning their average rating fell below your low-score threshold.
• Low NPS - "Employees who have provided a low score for the eNPS question," meaning they answered as detractors.

For each named employee, the list shows their participation rate, their engagement score, and their eNPS score. From there you can click into an individual to see more of their history.

The one thing to hold onto: every one of these lists is drawn only from confidential surveys. These lists are built only from surveys that were run as confidential, and the on-screen descriptions say the overview is "across all confidential surveys." An anonymous survey simply never enters this feature. So the named list isn't a way around anonymity; it's a separate capability that only ever operates on data you chose to collect with identity attached.

The thresholds behind each signal

The signals aren't arbitrary. Each one has a defined line, and it's worth knowing what they are so the lists make sense.

Note that the eNPS at-risk cut and the standard eNPS detractor band (0-6) sit very close together but can differ by a point at the boundary, so treat the "low eNPS" list as approximately your detractors rather than an exact match to the breakdown card. For the canonical detractor math, see What is eNPS and how is it calculated?.

Silent employees: the one signal that's identity-agnostic

Of the three signals, "silent" is special, and it's worth calling out. Whether someone submitted a response is participation data, not response content - and CultureMonkey treats those two things completely differently. Knowing that a person hasn't answered tells you nothing about what they would have said.

This is why you can always nudge non-responders even on an anonymous survey: the system targets the reminder at everyone without a submission, without ever showing you their answers (because there are none yet) or joining them to anyone else's. What stays restricted is the named roster of non-responders, which is identity data and is permission-gated rather than self-serve. The full treatment of this distinction lives in Identify and follow up with non-responders.

The practical upshot: a persistently silent employee is a real signal you can act on operationally (check that invites are reaching them, nudge them on the right channel) without ever touching the anonymity of their eventual answers.

A related signal: alerts on negative feedback

Separate from the three at-risk lists, CultureMonkey can also flag strongly negative open-text feedback as it arrives. When enabled, an incoming comment with sufficiently negative sentiment can notify your admins so a critical concern doesn't sit unread, and it can optionally send the employee an automatic acknowledgment. This is a sentiment signal on comments, not a low-score or low-eNPS signal, and it does not de-anonymize anything: on an anonymous survey the alert points to the comment and its theme, never to a person.

Where a survey is confidential, this can pair with a "conversation" flow that lets an admin or manager reach out on a specific piece of feedback. As always, that reach-out capability belongs to non-anonymous data and can be turned off for managers and sub-admins by account setting.

From signal to action

Finding at-risk people only matters if it changes something. The right next step depends on which mode you're in, and both paths are legitimate.

1. 1On anonymous surveys, act on the pattern. Take the low driver score, the detractor theme, or the struggling segment to the relevant leader. Build an action plan around the root cause, assign an owner, and close the loop with the group. Speak to themes, never to "whoever said this."
2. 2On confidential surveys, act on the person. Where you deliberately collected identity - a lifecycle check-in, an onboarding survey - use the named at-risk list to have a real, human conversation. Reach out with care, and remember the person knew this survey wasn't anonymous.
3. 3For silent employees, fix participation first. A nudge on the right channel, or diagnosing a delivery problem for a whole team, often does more than anything else. Silence can be a logistics issue, not a sentiment one.
4. 4Re-measure. Watch whether the low scores lift and the silent people re-engage over the next cycle. Movement is the proof your action worked.

Best practices

• Decide the mode before you launch, not after. If you might need to reach out to individuals, run the survey as confidential and tell people. Wishing an anonymous survey were identified after the fact is the one thing you cannot fix.
• Default to anonymous for engagement and eNPS. Honest sentiment needs anonymity. Reserve confidential mode for moments where personal follow-up is the explicit point.
• Read low scores alongside comments. A low number tells you where; the comments tell you why. Acting on the "where" without the "why" leads to the wrong fix.
• Trend beats snapshot. One low score is noise. The same person or team scoring low across cycles is signal. Watch the direction.
• Never stack filters to guess a name. It doesn't work (the product blocks small combinations) and it poisons trust. Treat anonymity as a promise you keep.

Frequently asked questions

Can I see who gave a low score on an anonymous survey?

No. On an anonymous survey, individual answers are never linked to a person, so there is no way for anyone - including an account admin - to see who scored low. You can see aggregate patterns (low driver scores, the detractor count, struggling segments) but never a name. This is enforced in the product, not just a policy.

So how does the named "at-risk employees" list exist at all?

It only ever draws from confidential (non-anonymous) surveys - ones you deliberately ran with identity attached. The underlying query filters strictly to surveys where anonymity is turned off. An anonymous survey never feeds that list. Named insight is a separate capability that operates only on data you chose to collect non-anonymously.

Is there a setting to reveal identities on an anonymous survey?

No. There is no toggle, export, or filter that de-anonymizes an anonymous survey. The only way to have individual-level insight is to run the survey as confidential from the start and communicate that to respondents.

What counts as a "low" score?

By default, a rating below 6 on a 0-10 scale, below 3 on a 5-point scale, or below 60% if you display scores as percentages. For eNPS, it's an answer in the detractor range. These follow whatever scale your account uses for reports.

Can I find silent employees without breaking anonymity?

Yes. Whether someone submitted is participation data, not response content, so you can always see how many people are silent and remind exactly those non-responders - without ever seeing their answers. A named roster of non-responders is permission-gated because it's identity data. See Identify and follow up with non-responders.

Why does the "low eNPS" list not exactly match my detractor count?

The at-risk cut and the standard detractor band (0-6) sit a point apart at the boundary, so the "low eNPS" list is approximately, not exactly, your detractors. For the precise eNPS math, see What is eNPS and how is it calculated?.

Where to go next

• Understand the protection that governs all of this: How anonymity keeps your feedback safe
• Chase silent employees the safe way: Identify and follow up with non-responders
• Turn a finding into change: Create an action plan
• Read the detractor split behind your eNPS: View the eNPS breakdown