The Cost of Not Knowing Your Judge

The most expensive miscalculation in litigation is rarely a misread of the law. It is arguing a strong case in a register the bench was never going to respond to, and paying for the effort twice.

Two counsel can take the same set of facts, the same authorities and the same client, and walk out of court with opposite results. Every experienced litigator has watched it happen, and most have quietly explained it away as the luck of the listing. It is a comfortable story, because luck is nobody's fault. It is also, for the most part, wrong. What looks like luck is very often information that one side had and the other did not: an understanding of how the particular judge in front of them tends to think, what they find persuasive, and what quietly irritates them.

The law is the part of litigation that firms research obsessively. The judge is the part they tend to guess at. That asymmetry is expensive, and the bill arrives in a form that never appears as a line item.

The judge is a variable, not a constant

We are trained to talk about the law as though it produces answers. Apply the right authority to the right facts and the outcome follows. If that were the whole picture, the identity of the judge would barely matter. The evidence says otherwise, and it says so loudly.

The clearest data comes from asylum adjudication, where large caseloads and effectively random assignment make judge-to-judge comparison possible. In the United States, a government review found that once you strip out the differences between cases and look at a representative applicant with identical characteristics, grant rates still swung by around thirty to forty percentage points from one court to another. Independent analysis of individual judges has been starker still. At one immigration court, the odds of a claim being refused ranged from roughly eleven per cent to nearly ninety-nine per cent depending only on which judge happened to be assigned. Researchers gave the phenomenon its own name: refugee roulette.

Britain does not publish the equivalent figures. There is no public dataset breaking down outcomes by individual judge here, which is itself part of the problem worth sitting with. What we do have points the same way. When the BBC examined asylum appeals across hearing centres, it found wide variation in success rates that could not be explained by the merits of the cases alone, prompting one solicitor to observe that two cases of equal strength could land completely differently depending on the judge and the venue.

Asylum is an extreme case, chosen because the numbers are visible. But nobody who has practised in any field believes the effect is confined to it. Commercial judges, family judges and first-instance district judges all bring settled habits of mind to the bench. The difference in ordinary civil work is not that judicial temperament stops mattering. It is that no one is measuring it, so the variation stays invisible and gets filed, once again, under luck.

What knowing your judge really means

There is an unhelpful version of this idea that needs setting aside first. Knowing your judge does not mean predicting the result, and it certainly does not mean gaming it. Any tool or practice that promised to tell you how a named judge will rule would be both professionally troubling and, frankly, dishonest about what published material can support.

The useful version is narrower and far more practical. It is about register. Judges, like all expert readers, have preferences in how an argument is built. Some want the authorities marshalled tightly and the skeleton kept lean. Some reason from principle and lose patience with a case that leads on the facts. Some carry a known scepticism about a particular line of argument, or a habit of returning to a favoured line of authority, or firm expectations about bundles, time estimates and procedural housekeeping that shape a hearing before the substance is even reached.

None of that is secret. It is written, in effect, across a judge's published judgments: which authorities they cite and how, how they structure reasoning, where they signal impatience, what kind of submission tends to persuade them and what tends to fall flat. An advocate who has absorbed that record arrives having pitched the argument in the register most likely to be heard. An advocate who has not is making a strong case in a language the room does not favour, and hoping.

That is the real content of the phrase. Not "will I win," but "am I arguing this the way this particular bench is most likely to receive well." The first question is unanswerable and improper. The second is preparation, and it is the kind the best advocates have always done by instinct.

Counting the cost properly

Here is where the invoice comes in, because the cost of not knowing your judge is entirely real. It is just distributed in a way that hides it.

Start with the price of an hour. The guideline hourly rates that courts in England and Wales use for assessing costs run, from January 2026, up to £579 an hour for a senior solicitor on heavy commercial work in London, and the private market climbs well past that, with large City firms commonly charging between roughly £500 and £900 an hour and the top of the market higher again. Every hour of preparation is an hour billed at rates like those. The question is not whether the hours are spent. It is whether they are spent on the things that move the particular judge.

Now picture the ordinary waste. A team spends two days building an argument around a line of authority the assigned judge has, in past judgments, treated coolly, when a different and equally available line would have landed. A skeleton is drafted long and discursive for a judge known to reward economy, and is quietly discounted before the hearing begins. An authorities bundle is assembled without regard to how this judge likes to be taken through the law, and has to be reworked. A time estimate is set without checking how this judge runs a list, and the matter goes part-heard, then adjourned. Each of these is hours at commercial rates, and none of them shows up on the bill as "cost of not knowing the judge." It shows up as ordinary work, which is precisely why it never gets questioned.

The strategic cost is larger and even harder to see. Arguing in the wrong register does not merely waste preparation. It spends credibility with the bench early, forfeits the momentum of a strong opening, and occasionally converts a winnable matter into a lost one. There is no timesheet for a case that should have settled favourably and did not, because the other side read the judge better. That is the most expensive hour of all, and it is invisible by definition.

There is a related signal worth noting. In the asylum data, applicants with legal representation were granted at up to three times the rate of those without. Representation is, at bottom, the ability to present a case in the form the tribunal can best act on. It is a blunt proxy, but it points at the same truth: how a case is put to a particular decision-maker measurably changes what happens to it.

Why this edge has stayed unevenly held

If understanding the judge is this valuable, why is it not standard practice, done well, everywhere? The answer is that the signal has always been buried and the work has always been manual.

The material exists, scattered across thousands of published judgments, but extracting a usable picture of how a given judge reasons has meant reading widely, remembering shrewdly, and then holding it all as tacit knowledge. In practice that knowledge concentrates in a few senior practitioners and a handful of sets who have appeared before the same judges for years. It is chambers lore, passed down unevenly and rarely written down. A junior in her second year has no realistic way to build it from scratch before a hearing, and no firm can afford to have its best people spend billable days reconstructing by hand what is, in principle, already on the public record.

So the edge goes to whoever happens to hold the institutional memory, and everyone else pays the cost of not knowing without ever seeing an alternative. That is an inefficiency, not a law of nature, and inefficiencies of exactly this shape are the ones technology tends to close.

From tacit knowledge to prepared advantage

This is the problem Probatur is built to solve, and it is worth being precise about how, because the how is what keeps it on the right side of both good sense and professional duty.

Probatur reads the public record at a scale no individual can. It draws out, from a judge's own published judgments, the patterns that make up their register: the authorities they lean on, the way they structure reasoning, the procedural expectations that tend to shape their hearings. It sits alongside semantic search across the case law and signals from the wider legal climate, so that the picture of the bench arrives together with the substance of the matter. And it presents all of it as preparation for the advocate's own judgment, a briefing to be verified against the primary record, never a prediction to be relied on and never a substitute for the professional's decision.

What that changes is who gets to hold the advantage. The understanding that used to take years of appearances, or a well-connected set, becomes available to a second-year solicitor the afternoon before a hearing. The days of manual digging that most teams cannot justify collapse into minutes. And the cost of not knowing your judge, the one that has been hiding inside ordinary preparation all along, starts to come down.

The bill you are already paying

The uncomfortable truth is that no litigator gets to choose whether the judge matters. That decision is made for you the moment the case is listed. The only choice is whether you prepare for the judge you actually have or the generic one you imagine.

Firms that treat the bench as a constant will keep paying the cost of not knowing it, in wasted hours, in blunted arguments and in the occasional winnable case that slips away, and they will keep filing all of it under bad luck. The firms that treat the judge as what they are, a variable you can study, will spend the same hours to better effect and win a little more often for reasons their opponents will struggle to name.

You are already paying the cost of not knowing your judge. It simply does not appear on the invoice. The only real question is whether you would rather keep paying it, or start recovering it.