Evidence from the record · What it means for leaders
The Post
Office
When the evidence became the enemy.
The best-documented institutional failure in Britain
Two cross-examined High Court judgments. A statutory public inquiry with the power to compel. Millions of disclosed internal documents, and years of testimony given under oath.
No other case in this series comes close to that standard of evidence. And unlike the others, the organisation here did not fail commercially at all. It is still trading. What failed was its capacity to believe anything that contradicted it.
That is why it belongs in a series about judgement rather than a series about insolvency. An organisation can be financially sound, operationally busy and completely unable to process information that threatens its own account of reality.
About the institution, not the people it harmed
Around a thousand people were prosecuted and convicted on the strength of data from this system. Roughly ten thousand are now eligible claimants for redress. The inquiry chair described the human impact as disastrous, and it was.
Those lives are not material for a leadership framework. What follows examines only how the organisation handled evidence—because that mechanism is transferable, and the suffering is not something to be mined for lessons.
Twenty years of warnings.
Thousands of people affected.
One system believed.
People likely prosecuted and convicted, on the inquiry’s estimate
Currently eligible claimants across the redress schemes
Years between the system’s rollout and the decisive judgments
Pages in the judgment that settled how the system actually worked
What happened
Twenty years in which the same objection was raised, over and over, by people with nothing to gain from raising it.
Horizon is rolled out
A new IT system is introduced across the Post Office branch network. Subpostmasters begin reporting unexplained accounting shortfalls. Under their contracts, they are held responsible for making good the missing money.
The same objection keeps returning
Subpostmasters repeatedly report that Horizon is generating unexplained discrepancies. They are told the system is robust and that nobody else is experiencing the same problem.
The cases begin to connect
Reporting brings individual accounts together. People who had each been told they were alone discover that the same pattern has occurred across the network.
An independent investigation begins
Forensic accountants are appointed to investigate concerns about Horizon and the treatment of subpostmasters. Their work identifies defects, weaknesses and failures in the organisation’s approach.
The institution defends its account
Evidence challenging Horizon accumulates, but the Post Office continues to maintain that the system is robust. The independent investigation and mediation process come to an end without resolving the underlying dispute.
The challenge reaches court
Hundreds of former subpostmasters bring group litigation against the Post Office. For the first time, the organisation’s claims about Horizon are tested through disclosure, expert evidence and cross-examination.
The judgments dismantle the story
Two High Court judgments reject central parts of the Post Office’s account. Horizon is found capable of producing discrepancies, and the contractual relationship with subpostmasters is found to have been oppressive.
Convictions are overturned
The Court of Appeal quashes dozens of convictions, finding that the Post Office’s failures made the prosecutions an affront to the public conscience.
The public inquiry examines the institution
Witnesses give evidence under oath and millions of internal documents are disclosed. The inquiry examines not only what Horizon did, but how the Post Office responded to information that contradicted its position.
The inquiry’s first report lands
The inquiry concludes that the Post Office maintained its belief in Horizon despite the accumulating evidence, with disastrous consequences for the people affected.
What the judgments and inquiry
said about the institution.
The Post Office maintained
its belief in Horizon despite
the accumulating evidence.
The inquiry found that concerns about Horizon were repeatedly discounted while confidence in the system was preserved.
A pattern, not an incident
The judgments describe a culture of excessive secrecy, a settled pattern of defensiveness, and a lack of transparency running through the organisation’s conduct of the matter.
This was not one bad decision made under pressure. It was a consistent posture, maintained by successive people who mostly inherited it rather than chose it—which is precisely what makes it worth studying.
A “malignant culture”
The inquiry identified culture, rather than any single decision or individual, as the main driver of what happened.
That is an uncomfortable conclusion, because culture is the thing every organisation believes it has a good one of. Nobody inside a malignant culture experiences it as malignant. They experience it as loyalty, rigour, and defending the organisation against people trying to damage it.
Two kinds of evidence.
Two standards of proof.
The organisation was never short of information. It had, from very early on, two streams of it—and they disagreed.
What determined the outcome was not the quality of either stream. It was that the two were held to completely different standards before they would be believed.
A shortfall in this branch
One source. Untested.
The same fault, reported independently, hundreds of times
Many sources. Repeated evidence.
The stronger evidence was the one the organisation refused to treat as evidence.
What the signal actually looked like
Hundreds of independent accounts
Statistically, the subpostmasters’ account was far the better evidence. Hundreds of people who had never met, in unconnected branches, independently describing the same behaviour, is close to the definition of a reliable signal.
A single unaudited system output is not.
The organisation had already decided which one it believed.
It only ever asked whether the other could be discredited.
Why more evidence
made it worse.
In the other studies in this series, the failure was that a signal did not arrive, or arrived and produced nothing. Here something stranger happened. The signal arrived repeatedly, and each arrival strengthened the defence.
Once an organisation has acted on a belief—prosecuted someone, published a statement, defended a position in court—new contrary evidence no longer threatens only the belief. It threatens everything already done in its name.
Each defence raised the cost of the next admission
Every prosecution made it more expensive to accept that the system was unreliable. Every public assurance made it harder to correct. Every court victory became a precedent to protect.
By the end, accepting the evidence would have meant accepting that the institution had ruined hundreds of innocent people. That is not a conclusion an organisation reasons its way to voluntarily—and the longer it waited, the more true that became.
Attacking the source rather than testing the claim
The tell is not disagreement. It is where the organisation’s energy goes: applying to strike out evidence, seeking to remove a judge, and closing down an investigation it had itself commissioned.
All of that effort was directed at the standing of the people making the claim. Almost none of it was directed at whether the claim was true—which could have been settled, at any point in twenty years, by a proper independent examination of the system.
What was missing
The Shortboard model describes twelve attributes an organisation builds across three cumulative waves.
Wave One is Lean Dependability—the ability to do what you say you will do, honestly and without drama. Wave Two enables scalable growth. Wave Three creates perpetual relevance.
Capabilities in the later waves cannot hold when Wave One is hollow, because there is nothing underneath them to take the weight.
Three Wave One attributes are absent here. The fourth is marked differently, and the reason is important: this organisation was never subject to the market force at all.
Lean Dependability
Scrappy resourcefulness
Using limited resources deliberately while understanding and evidencing how the operation performs.
Healthy
conflict
Surfacing disagreement early and allowing evidence to challenge authority before decisions are made.
Ruthless consolidation
Choosing what to stop so attention, money and capacity remain focused on what matters most.
Distributed ownership
Placing authority and accountability close to the people with the knowledge required to act.
New Growth
Continuous reinvention
Continually renewing propositions instead of treating current success as permanent.
Structural fluidity
Reconfiguring people and resources rapidly around emerging priorities and opportunities.
Strategic optionality
Maintaining several credible paths rather than committing too early to a single answer.
Platform thinking
Building reusable capabilities that make future growth faster, easier and less expensive.
Perpetual Relevance
Pioneer sanctuaries
Protecting emerging ideas from the demands and assumptions of the established organisation.
The awe-driven mindset
Remaining curious about what is becoming possible instead of defending what is already known.
The pioneer’s leap
Acting decisively when evidence is incomplete but the cost of waiting is greater.
Supply-driven optionality
Creating new possibilities from emerging capabilities before established demand is visible.
No customer could leave, so nothing forced a correction
In a competitive business, a twenty-year dispute with your own branch network would show up in market share long before it showed up in court. Here it could not. Subpostmasters could not take their branch elsewhere, and customers had no visibility of the argument at all.
The only correcting mechanism available was litigation—slow, ruinously expensive and asymmetric in favour of the institution. It eventually worked. It took twenty years.
The gaps, and the one
that never applied.
Each of these assessments rests on findings made in court or by the statutory inquiry.
Healthy conflict
The dominant failure by a distance. Challenge was not merely unwelcome; it was litigated against. An organisation that applies to strike out evidence it disagrees with, and then applies to remove the judge who heard it, has passed well beyond a reluctance to be contradicted.
Distributed ownership
The people with the clearest view of the problem were the furthest from any authority to act on it. Subpostmasters had no standing at all in the organisation’s account of itself. Internally, what was known in one part did not reliably reach the people answering for it in another.
Scrappy resourcefulness
The organisation could not, or would not, examine the reliability of its own operational data. Whether the system could be wrong was answerable throughout—by testing it. That test was avoided for two decades and was ultimately performed by a court.
Ruthless consolidation
Marked as untested rather than absent. This is the market-facing attribute, and no market force ever reached this organisation. That is not a credit to it. It means one of the four forces that normally corrects an institution was simply not connected.
Worth carrying into your own organisation: which of the four forces actually reaches you?
Monopolies, internal service functions, regulated bodies and heavily contracted suppliers all share this exposure.
Where customers cannot leave, the correction has to be built deliberately, because nothing external will supply it.
Still being resolved
This is the only study in the series where the record is not yet complete.
Volume 1 of the inquiry’s final report was published in July 2025, dealing with human impact and redress and making nineteen recommendations. The chair concluded that the compensation schemes had not, in many cases, delivered full, fair and prompt redress.
The remaining volumes—the ones addressing how this happened and who was responsible—are still being prepared and will be published together. That means this study will need revisiting.
The litigation settles
Much of the settlement is consumed by the cost of having brought the case.
Convictions are quashed
Convictions begin to be overturned. The statutory inquiry gains its powers.
Public attention arrives
A television dramatisation achieves in weeks what two decades of evidence had not.
The process continues
Volume 1 reports. Redress schemes continue. Further inquiry volumes are awaited.
The third panel deserves attention.
The facts were established in a 313-page public judgment in 2019. Widespread political and public urgency arrived in 2024, after a drama series.
The evidence had been available, free and definitive, for five years.
Being right is not a mechanism
The subpostmasters were right from the beginning. Being right produced no outcome for twenty years, and the thing that finally moved the institution was neither evidence nor argument but attention.
If your organisation has someone who has been right and ignored for years, you do not have a communication problem. You have a structural one, and it will not resolve itself by their being even more right.
Decide now which one
you would believe.
When your system and your people disagree, you have already decided which one you trust.
That decision was made long before the disagreement, by how the organisation is built: who is believed by default, what it costs to contradict the official account, and whether anyone has the standing to demand that the system itself be tested.
This is not a historical curiosity. Every organisation is now installing systems whose outputs it does not fully understand, and every one of them will eventually produce a number that a competent, honest person on the front line says is wrong.
When our data contradicts what our frontline is telling us, which one do we investigate first?
What does it cost someone here to be the first to say the system is wrong?
Who inside this organisation benefits from that system being considered reliable?
Count the independent reports
When several people who have no contact with each other describe the same fault, that is not a coordinated complaint. It is the strongest evidence you will ever get for free.
Treat the second independent report of anything as a finding, not as a second complaint—and give someone the explicit job of noticing when it happens.
How we know this,
and what we don’t.
The strongest record in this series, and the only one that is still incomplete.
Two substantive High Court judgments, following trials in which the evidence was tested under cross-examination rather than asserted.
A statutory public inquiry with powers to compel documents and testimony, conducted in public over three years, with the material published as it went.
A Court of Appeal ruling on the safety of the convictions. This is as close to established fact as any organisational failure gets.
The volumes addressing how this happened, and who bears responsibility, have not yet been published. Conclusions about individual conduct are not yet settled.
Litigation evidence is shaped by what each side chose to contest, and the inquiry’s narrative was built after the outcome was known.
This case is also now heavily mediated by dramatisation, which makes the popular version easy to mistake for the record.
Structure, not villainy
It is tempting to read this as a story about bad people, and the remaining volumes may well make findings about individuals. But the posture described in the judgments persisted across two decades and several leadership teams.
Treat it as evidence of what an institution does to the people inside it—not as proof that this particular institution was uniquely staffed by people unlike your own.
Bates and others v Post Office Ltd, Judgment No. 3 (Common Issues), March 2019, and Judgment No. 6 (Horizon Issues), December 2019, High Court · Court of Appeal (Criminal Division), April 2021 · Post Office Horizon IT Inquiry, Volume 1 of the final report, July 2025, and published hearing evidence.
Subscribe to Notes from the Water
Our weekly newsletter brings readers the latest insights, studies, and observations relating to institutional judgement and the business design