Nobody Thought There Were Barriers
Nobody Thought There Were Barriers
What an audit of raising concerns found, and the question it couldn’t ask
In March 2026, the Northern Ireland Audit Office published a review of how concerns are raised across the public sector: whistleblowing arrangements, designated officers, policies, and case numbers. It is a careful, fair-minded report, and its conclusion is that the formal machinery has improved considerably while the culture has not.
Buried in Part Four is a sentence that explains more than the report does.
“During our engagement, none of those responsible for raising concerns in public bodies highlighted any significant barriers that they felt were preventing people speaking up.”
The auditors asked the people who run the system whether the system has barriers. The people who run the system said no.
A few lines later, the report notes that several bodies had received very few concerns over the five-year period.
Two readings of an absence
This is the problem of this subject in miniature:
An organisation with almost no concerns raised has two explanations available. The first is that there is nothing to raise. The second is that raising any concerns is expensive enough that people don’t. Both produce identical data. Silence looks the same either way.
A case from a different field shows how badly this can go.
In 2008, hundreds of small-scale diamond miners in Zimbabwe were killed by government forces, and an international ban followed. In 2011 inspectors from the Kimberley Process visited, found no ongoing human rights abuses, and the ban was lifted.
The torture camps were still standing. A local resident, asked what had changed, said nothing had. The regime had simply become efficient enough that very few people now risked the attention that would produce violence.
The absence of the event was not the absence of the mechanism. It was evidence that the mechanism was working.
The inspectors were not careless. They looked for what they had come to find, found none of it, and reported accurately. What they could not see was the thing producing the quiet, and the people who could see it had the strongest possible reason not to tell a visiting delegation.
An organisation receiving no concerns is in the same position. The silence is data, and it does not interpret itself. Whether it means there is nothing wrong or that raising is expensive depends on information the organisation does not hold because the people who hold it are the ones who decided not to speak.
Which is why the question put to the designated officers could not have produced a useful answer. The people running a reporting system are the same people a potential raiser assesses before deciding whether to use it. Asking them whether the route is safe is asking the route about itself.
The information needed is held by the people who didn’t come forward, and by definition, they aren’t in the room.
What the data says instead
The report does contain the answer, in two places.
The first is the uphold rate. Across five years and 519 completed investigations, 300 were not upheld. Only about one in ten concerns was upheld; in central government the figure is nine per cent.
From the position of someone deciding whether to raise something, that is roughly a nine-in-ten chance that nothing changes, plus whatever follows for them personally. Not a calculation requiring cynicism. Arithmetic, available to anyone who has watched a colleague try.
The second is sharper, and I would put it on the front page.
In four of the last five years, the majority of concerns came from people outside the organisation. The single exception was the year the raising-concerns framework was published.
Consider what that comparison isolates.
External people have less information. They cannot see what staff see, they do not know the history, and they are often guessing. They also have no legal protection at all – the Public Interest Disclosure legislation covers workers, not members of the public.
Staff have more information and statutory protection.
Staff raise fewer concerns.
The difference between the two groups is not knowledge, and it is not protection on paper. It is that one group has to go on working there afterwards.
That is as close to a controlled comparison as this subject ever gets, and it points in one direction.
The case everyone already knows about
The report includes a third thing, but it is not presented as evidence of anything.
Dr Bronckaers, a veterinary officer in the Department of Agriculture, Environment and Rural Affairs, raised concerns about animal welfare and about the effect of deleted cattle movements on traceability. An independent report later found that her concerns had not been treated appropriately. She received a £1.25 million settlement and an unreserved apology.
The report cites this as background to why the framework exists. Fair enough.
But consider what that case is, from the position of a civil servant weighing whether to put something in writing. It is the most visible and most specific information available about what raising a concern costs. Not because she lost, but because she won, and it took years and a legal process to get there.
Nobody needs to be told the lesson. The settlement is the lesson, and it has circulated more widely than any policy document.
The thing none of this measures
What the report measures is capability and awareness: whether policies exist, whether people know how to use them, whether officers are trained.
On that, the findings are unflattering. Designated officers in the largest public bodies had received little or no recent training for the role. Speak Up Champions, required by the framework, largely do not exist: several departmental policies state the role will be performed by the designated officer, the same person. No organisation has a timescale for responding, and some cases have remained open for years.
All of which is worth fixing, and none of which is the constraint.
Because what the report cannot measure, and what every potential raiser is assessing, is what happened to the last person who did it.
That is not an oversight. It is genuinely hard to measure, and it is nobody’s obligation to record. But it is the only variable that matters at the moment of decision, and the system has no instrument for it.
Which is why the recommendations stop short
The report makes eight recommendations. Appoint champions. Provide training. Measure awareness. Promote a positive culture, led from the top.
All sensible, and most will be implemented, because they are the kind of thing an organisation can do.
But consider what each asks for. Training raises the quality of handling. Champions create a role. Awareness measurement tells you whether people know the route exists.
Not one of them changes what using the route costs.
What would
Three things, none of them a program.
Measure what the system currently cannot see.
Not whether a policy exists, but what happened afterwards to the last several people who used it. Did they remain in post? Did they move sideways? What did their next appraisal say? That information sits in HR systems already, and nobody has assembled it. A board that saw it once would not need telling what its reporting rate means.
Treat the internal-external ratio as the headline figure.
It is already being collected. When external concerns outnumber internal ones, the people with the most information and the most protection raise the least, and that comparison needs no interpretation.
And separate the receiving function from the organisation’s interest.
The designated officer sits inside the body the concern is about, frequently in HR, sometimes conducting the investigation personally. That is a function asked to supply independence it does not structurally have, and no amount of training alters what the raiser is reading when they decide whether to approach it.
The finding underneath
The report asked sensible questions of the right people and reached a fair conclusion about the state of the machinery.
The one question it could not ask was of the people who considered raising something and decided not to. Their reasoning is the whole subject, and it is held entirely outside the system trying to measure itself.
Which leaves the most reliable instrument as the one the report has already published and not recognised: the ratio between those who have to go on working there and those who don’t.