Why does every fee earner do the same job differently?

Where does the variation actually come from?

Take a firm of forty people and ask five fee earners to produce the same deliverable: a client update, a scope note, a fee proposal. You will get five documents with different headings, in a different order, at different levels of detail. None of them is wrong. Each took a different amount of time to write. And nobody has ever put the five side by side, so nobody has ever decided which one is the firm's way.

The variation was not designed. It accumulated. Someone arrived from another practice and kept their old format. Someone built a spreadsheet during a busy quarter and is still using it four years later. A partner dislikes tables, so their team stopped using tables. That is the drift described in why templates rot, except it happens to behaviour as well as to files.

Nobody intervened because nothing visibly broke. The work went out, the client paid, the file was closed. Variation never appears as a line anywhere, so it never gets challenged.

Is it professional judgement or just habit?

Raise it and the defence is immediate and sincere: we are a professional firm, our people exercise judgement, we are not a production line. True about content. Which risk to flag, how hard to push a point, what a particular client can absorb: all judgement, and the reason the firm can charge what it charges.

But look at where the variation sits. In the order of sections. In whether assumptions are stated. In whether the summary comes first or last. In who reads it before it goes out, and whether anyone does. In whether it is saved to the matter file, the shared drive or someone's sent items. None of that is judgement about the client. It is habit that has never been asked to justify itself.

There is a simple test. Ask why they do it their way. If the answer is about the client, the matter or the risk, it is judgement, and you should leave it alone. If the answer is that it is how they were shown when they joined, it is habit. Firms conflate the two constantly, and the conflation protects a lot of waste.

Why does this only become visible when you try to automate?

You can run a firm for years on undocumented variation. You cannot rebuild a job on it. The first week of any rebuild is mapping: sitting with the people who do the work and tracing every input, template, handoff and check. That is the week the firm discovers it does not have one process. It has as many as it has people doing the job.

Three people give three answers to what happens next. A fourth mentions a check the others have never heard of. Someone produces a reference document the rest of the team did not know existed. This is the most useful thing the mapping does, and it is why documenting a process is worth doing even if you never automate anything.

It also explains why projects stall. Software cannot settle a disagreement about how work should be done. It encodes whichever version it was shown, and the people who do it differently carry on quietly. That is the mechanism behind most of the failures in why AI rollouts fail.

What does the variation cost before anyone automates anything?

It is expensive long before any technology question arises.

• Review takes longer. A reviewer reading an unfamiliar shape has to work out the structure before assessing the substance.
• Rework rises. With no agreed standard, comments are preference rather than correction, and drafts go round again. That loop is the cost of rework.
• Juniors learn nine ways. A trainee working across four supervisors learns four house styles and no firm style.
• Knowledge walks out. If the method lives in one person's habits, it leaves with them, which is what makes a departure so disruptive.
• Clients notice. A client with two matters open sees two standards of reporting from one firm.

None of this appears in management information. It shows up as capacity pressure, missed deadlines and partners working weekends, which get blamed on volume.

Which version should become the firm's way?

Not the most senior person's by default, and not an average of all of them, because you cannot average a document into existence. Pick a real example and improve it.

Useful criteria: the version that comes back from review with the fewest comments; the version a client has praised; the version that covers what a regulator or an insurer would want on the file; the version that takes least time to produce without losing anything. Where two point the same way, you have your answer. Where they conflict, that is the conversation the partners should be having, and it takes an hour, not a quarter.

Then write it down in a form people will use. Not a policy nobody opens, but the actual shape of the deliverable, which is what preparing templates for automation covers, and the actual sequence of steps, which is what standard operating procedures are for.

Standardise the container, not the thinking. The shape, the sequence and the checks are the firm's. What goes inside stays with the fee earner.

Does every exception have to be covered?

No, and trying is how this turns into a programme that never finishes. Most deliverables follow a common path most of the time, with a handful of genuine exceptions that need a person to think. Agree the common path, name the exceptions, and let a fee earner handle them, which is the right use of their time.

The common path is what gets rebuilt. Client reporting in a surveying practice, scope documents in an agency, the file note in a law firm: predictable shape, variable content. The rebuild is tested on live work rather than demonstrated on a sample, so the awkward cases surface while there is still time to decide what the process does with them.

What should a firm look at first?

Pick one deliverable that leaves the firm regularly. Ask four people to send you the last one they produced, unedited. Read them next to each other. The differences will be obvious within ten minutes, and so will the question of which one you would be happy to call the standard.

Then ask how long each took. That gap is the cost of having no agreed method, and it compounds across every fee earner and every week of the year. The unbillable hours calculator turns it into a yearly figure using your own charge-out rate, and the process priority scorer helps you judge whether this is the job to fix first. The audit takes about three minutes and names the process to tackle.

The firms that get anywhere with this are not the ones with the best software. They are the ones willing to decide, once, how a job is done.