Somebody presses send on the survey. Two weeks pass. The results land in a leadership meeting on a Tuesday afternoon.
The room goes quiet, because the numbers are worse than anyone expected. Psychological safety has come out low. Trust in senior leadership has come out lower. Someone has written something in the free text box that will be quoted back for months.
This is one of the more uncomfortable moments in a growing company. It is also one of the most useful. Which of the two it turns out to be depends almost entirely on what the leadership team does in the fortnight that follows.
The first instinct is to argue with the survey
The conversation usually starts with the survey rather than the findings. Was the sample big enough? Did people understand the questions? Was it run in the week of the reorganisation, or just after the bonus announcement? Are we comparing ourselves against the right benchmark? Did that one team answer honestly, or did they treat it as a chance to make a point?
Some of those questions are fair. Timing does affect results. A badly worded question does produce a bad answer.
The problem is what happens next. The leadership team spends its energy on the method and never reaches the finding. Three months later the survey has been rewritten and the benchmark has been changed. Nothing about how anyone works has moved.
There is a simpler way to read it. You asked your people a question and they answered it. You have paid for information about problems you were going to meet anyway.
Why psychological safety scores drop as a company grows
A low psychological safety score is usually the first to appear as a company scales. It is worth understanding why, because it is rarely about anyone being unkind.
The term comes from Amy Edmondson's work at Harvard, set out in The Fearless Organization. It means something narrower than the way it tends to get used internally. It is about whether people believe they can raise a problem, admit a mistake or disagree with someone senior without it costing them. Google went looking for what separated its most effective teams and landed in the same place. It published what it found.
In a company of thirty or forty people, everybody can see how the founders handle a hard moment. Something goes wrong, people watch what happens, and they learn from it directly. The person who admits a mistake in front of the room and gets thanked for it teaches everyone else what is safe here. No policy required.
Past a hundred people that stops working, and it happens at fairly predictable sizes. Most people no longer experience the business through the founders. They experience it through their manager. And that manager was promoted because they were excellent at a different job. Nobody has shown them how to react when someone brings them bad news at five o'clock on a Friday.
So the reactions become inconsistent. Some teams stay safe. Others quietly stop speaking up. The average of those two things is the number on your slide.
Our founder, Terry Russell, has spent twenty five years in leadership teams doing this work. He puts it like this: as a company grows, process starts doing the job that trust used to. A psychological safety score falling is often the first visible sign of that handover going badly.
We have written about what breaks at each of those sizes in culture at 50 people vs culture at 500. The managers carrying all this were usually promoted for being excellent at something else. That is why hiring and promoting against behaviours matters more the bigger you get.
What to do when engagement survey results come back badly
There are two ways out of that room. The choice between them matters more than the score.
On the first path, the results get managed. The presentation goes to the board with the difficult slide moved to the appendix. A working group is formed. An engagement survey action plan with eleven workstreams is written, circulated, and never mentioned again. The following year the survey runs with different questions, the score improves slightly, and everyone agrees progress has been made. What people learn from this is that telling the truth in a survey produces a deck.
On the second path, the results get used. In the first week, somebody senior says out loud what the survey said and what was uncomfortable about it. One thing gets chosen. It changes visibly, and quickly enough that people connect the change to the survey they filled in. What people learn from this is that speaking up here leads to something happening. That is the only thing that will make them do it again next year.
The second path is harder in the first month. It is considerably easier in every month after that.
Your first two weeks after the results land
Speed matters more than polish. If the results come back on a Tuesday, people should hear something within a week, and it should come from a person rather than a slide deck.
That message needs to do three things.
It needs to say what the survey said, including the part that stings. Summarising your way around it tells everyone the summary cannot be trusted.
It needs to name one thing that is going to change. Choose it because it is achievable, not because it is the biggest number on the chart.
And it needs to be honest about what is not changing yet, and why. A list of eleven promises is a list of eleven future disappointments.
Then the change has to happen, and people have to be able to see it. This is the same distinction as the one between a value and a behaviour. A value is a word on a wall. A behaviour is something a person can be seen doing, which means it can be coached and recruited for. We have written about that in the laminated values problem.
A leadership team saying it will be more open is not observable. Publishing the decisions it makes each month is. Holding a session where the questions come unfiltered is. Telling people why a project was cancelled when saying nothing would have been easier is.
Terry's argument is that culture gets built through thousands of small interactions. In each one, leaders teach people three things: what matters here, what gets rewarded here, and what gets tolerated here. Your response to a bad survey is one of those interactions, and it is a loud one.
Surveys are one channel of employee voice among several. The CIPD's guidance on voice makes the same point about what happens when a channel exists and nothing follows it.
The three responses that cost the most trust
A few reactions reliably make things worse. They are worth naming, because none of them feels unreasonable at the time.
Hunting for the source. Working out which team gave the low scores, and having a quiet word with their manager, guarantees the next survey tells you nothing at all.
Explaining the score. Responding with context, reasons and a reminder of how hard the year has been asks your people to feel sympathy for the people with the power. That lands exactly as badly as it sounds.
Promising everything. Acting on one item properly builds more credibility than acknowledging fifteen.
Common questions
What does a low psychological safety score actually mean? It means enough people do not believe they can raise a problem, admit a mistake or disagree with someone senior without paying for it. It measures what people expect to happen to them. So it tells you about the reactions they have already seen, rather than about how anyone feels in the abstract.
What should you do after a bad employee engagement survey? Say what the results said within a week, including the uncomfortable part. Name one thing you are going to change. Be honest about what is not changing yet. Then make that one change visible quickly enough that people connect it to the survey they filled in.
How quickly should leadership respond to survey results? Within a week for the acknowledgement. Within a month for the first visible change. The gap between the survey closing and anyone hearing anything is what people read as the real answer.
Why do psychological safety scores drop as a company grows? Because the informal signals that taught people what was safe stop reaching everyone. Past roughly a hundred people, most people experience the business through their manager rather than the founders. Managers promoted for technical excellence have rarely been shown how to react when someone brings them bad news.
The question worth taking into the room
A bad first survey is a company telling its leadership team something in the only way it currently feels able to. That is worth knowing, even when it is unpleasant to read.
Terry tends to leave leadership teams with one question after a session on this, and it works just as well the week the results land. What is the one behaviour we need every new joiner to copy, and where specifically will they see it? If the answer is not obvious to the people around the table, that is the work. The survey has just told you it is overdue.
Terry ran a full online session on the three sizes where this tends to give way, at around 50, 150 and 500 people, and on what to put in place at each one – you can claim the accompanying handbook for free here. This gives you an exercise for working out what your survey results are actually telling you, and where to start.
