From “high, medium, low” to numbers

A client does not want a lecture on the law. A client wants to decide something: settle or go to court, sign or walk away, spend money on a fix or accept the problem. To decide, they need two things from you. How likely is it? And how bad could it be?

The traditional answer is a careful paragraph that ends: “there is a reasonable prospect of success, but the outcome is uncertain.” It is true, and it is useless. A client cannot add it to a budget, compare it with another problem, or explain it to a board.

In this course a risk has three parts, and you should be able to name all three for any problem you meet:

• An event that might happen. Be specific: “the court finds the contract was validly terminated”, not “legal trouble”.
• A chance that it happens: a number between 0 and 100 per cent.
• A consequence if it does: usually money, but also time, a lost relationship, or a stopped business.

Our running client is Alderwick Exports, a fictional mid-sized company. Its first problem is a dispute with a buyer who refused to pay for a shipment. Alderwick claims $200,000. We will come back to Alderwick in every module.

In 1964 an intelligence analyst called Sherman Kent noticed a problem. His colleagues wrote reports saying an event was “probable” or “possible”, and readers took very different things from the same word. He proposed fixed ranges: “almost certain” for about 93 per cent, “probable” for about 75, “chances about even” for 50, “probably not” for about 30, and “almost certainly not” for about 7.

You do not have to adopt his scale. The point is the problem he found: the same word means different numbers to different people, including to the same person on different days. Lawyers are not exempt.

Suppose the other side’s lawyer writes to Alderwick: “your claim is likely to succeed.” Alderwick could win $200,000. A trial would cost another $40,000. The buyer offers $90,000 now.

Is trial better than the offer? It depends on the chance of winning. The break-even is the chance at which the two choices are equal: (offer + trial cost) divided by the award, which is ($90,000 + $40,000) / $200,000 = 65 per cent. Above 65, go to trial. Below it, settle.

Now a reader who hears “likely” as 55 per cent settles. A reader who hears it as 85 per cent goes to trial. Same sentence, opposite decisions. The lab below shows this with your own numbers.

The fix is not to replace “likely” with “63 per cent”. A single figure sounds more exact than any lawyer’s judgement can be. Give a range and a reason: “we put the chance of success at 55 to 70 per cent, because the contract is clear but the key witness is weak.”

This keeps the plain-language feel, tells the client how unsure you are, and gives a number that can be used in the next steps of this course: expected value, decision trees and simulation.

• Using a loose word and assuming the reader hears what you meant.
• Giving one precise-looking figure with no range and no reason.
• Treating a number as a prediction. It is a way to state and test judgement, not a fact about the future.
• Chances that do not add up: for outcomes that cannot happen together and cover everything, they must total 100 per cent.
• Ranking risks only as high, medium and low, which hides differences of a hundred times inside each label.