When Both Stories Fit the Facts

A case study in structured evidence analysis — Part 1 of 3

This is the first article in a three-part series introducing the Structured Evidence Analysis Matrix (SEAM), a method I developed for testing competing accounts against the documentary record in contested legal matters.

Consider a person accused of stealing a car. They hold a driver’s licence. They knew where the car was kept. They had access to a key.

Every one of those facts is consistent with guilt. Every one is also consistent with innocence. Individually and collectively, they prove nothing, because they fit both stories equally well.

Now suppose that same person can be placed in another country on the day of the theft. That single fact does not merely weaken the accusation. It defeats it.

This is the insight at the heart of a discipline I have spent much of my career applying: evidence earns its weight not by fitting an account, but by being difficult to reconcile with the alternatives. Most contested litigation gets this backwards — and it costs people dearly.

The engagement

Recently I helped a party to a long-running, bitterly contested court matter prepare their evidence for a final hearing. I won’t identify the matter, the jurisdiction, or anyone involved, and I have altered or generalised details throughout this series. What matters is the shape of the problem, because it is a shape I see constantly.

The client faced serious cross-allegations. Each side had sworn accounts. The documentary record ran to thousands of pages — years of correspondence, more than ten thousand messages, financial records, third-party letters and reports, transcripts, and a history of earlier proceedings between the parties. Both sides could point to evidence that fitted their story.

That last sentence is the trap. When the client first walked me through the matter, they did what almost everyone does: they catalogued all the facts that supported their account. There were many. The problem was that the other side could do exactly the same thing, and had. Two narratives, each internally coherent, each garlanded with supporting facts, each asking the court to believe it.

A decision-maker confronted with that record has an almost impossible cognitive task. Nobody can hold ten thousand messages in their head. What actually happens is that humans — judges included, because judges are human — reach for a narrative that feels coherent and then, without ever intending to, notice the evidence that fits it. Psychologists call this confirmation bias. It is not a character flaw; it is the default setting of the human mind under information overload.

Most advocacy failures are not failures of merit. They are failures of clarity, sitting on a foundation of untested assumptions.

The wrong question and the right one

The client’s instinct — how much evidence supports my account? — is the wrong question. Consistency is cheap. As the car thief shows, a mountain of consistent facts can prove precisely nothing.

The right question is: which account is least contradicted by the record?

That reframing comes from intelligence analysis. In 1999 the CIA’s Center for the Study of Intelligence published Richards J. Heuer Jr’s Psychology of Intelligence Analysis, which introduced a technique called the Analysis of Competing Hypotheses (ACH). It remains one of the most widely taught structured analytic techniques in the intelligence world, in Australia and internationally. Its purpose is to defeat confirmation bias by forcing the analyst to test everyplausible explanation against the whole body of evidence — and to hunt for disconfirmation rather than support.

ACH is a superb method. But it was built for intelligence assessments, not courtrooms. Judges are not intelligence analysts, evidence in litigation must be traceable to documents that are actually before the court, and an analysis prepared by one party will rightly be met with scepticism unless its methodology can be independently verified.

So I built an adaptation designed specifically for legal disputes. I call it the Structured Evidence Analysis Matrix (SEAM). It keeps the analytical engine of ACH and adds the things a legal setting demands: accounts framed the way courts actually encounter them, a scoring scale that recognises material inconsistency, full traceability of every entry to a dated primary document, an express statement of the method’s limits, and — critically — a framework for independent expert validation of the methodology itself. The later articles in this series unpack each of these.

What we did

For this matter, the SEAM tested three competing accounts. Account A was the other side’s story: that the client had done what was alleged. Account C was the client’s story: that the reverse was true. And between them sat Account B — the genuinely neutral account that neither party’s allegations were made out, and the whole sorry record reflected nothing more than ordinary conflict, mistake, and miscommunication between parties locked in a broader dispute.

Including that neutral account matters enormously, and so does including the account adverse to the person preparing the matrix. A hypothesis set built only from stories you like is a straw-man exercise. A hypothesis set that includes “my client is wrong” and “nobody is right” is an analytical discipline.

We then distilled the sprawling record into seventeen evidentiary items — contemporaneous messages, financial records, the timing and fate of earlier applications, conduct records, and the internal features of the allegations themselves — and scored every item against every account on a four-level scale: Consistent, Equivocal, Inconsistent, or Materially inconsistent.

The matrix is read down the columns, not across the rows. A column full of “consistent” cells earns little; consistency, remember, is cheap. The analytical work is done by the inconsistencies.

What the matrix showed

When the scoring was complete, the pattern was stark. One account was contradicted by the contemporaneous record on nine of the seventeen items, including one material inconsistency — a sworn claim that the parties’ own correspondence flatly contradicted. The neutral account was contradicted three times and could not accommodate several of the record’s most significant features. And one account was not contradicted by a single item.

Seventeen items. Three accounts. One page of results a decision-maker can absorb in minutes, with every cell traceable to a dated document in the filed material.

I want to be careful here, because the method demands it. A SEAM does not decide anything, and it does not pretend to. It does not determine credibility, motive, or the ultimate facts — those are matters for the court, and the matrix says so on its face. What it does is change the shape of the conversation. Instead of two stories shouting past each other across thousands of pages, the court received a single transparent analysis showing which account survived contact with the record — an analysis it could interrogate, challenge, or re-score for itself.

For the client, the change was just as profound. The fog of “there’s so much evidence, where do I start?” resolved into something structured, defensible, and calm.

Next in this series

In Part 2, I open the bonnet: how ACH actually works, what diagnosticity means, and precisely what SEAM adds to make the technique fit for legal use. In Part 3, I cover the element that turns a clever analysis into court-ready material: the independent expert witness framework, the verification annexes, and the letter of instruction that holds it all together.

If you are a lawyer, advocate, or party facing a contested record that has outgrown narrative submissions, this is the work I do — independent, fixed-fee, and upstream of legal advice. You can read about my independent analysis services here.

Julian Talbot is the author of the Security Risk Management Body of Knowledge (SRMBOK) and has more than 35 years’ experience in risk management and analysis across government and regulated industries. He provides independent analytical reports for law firms, advocacy organisations, and parties to complex disputes. This series describes analytical methodology; it is not legal advice.

Subscribe to see the next in the series.