What is the bus factor of your accounting firm?

Engineers have a blunt measure called the bus factor: how many people would have to disappear before a project stalls. The idea travels well. The bus factor of an accounting firm is not one number but a count, task by task, of how many people can actually do the work today. Most firms of 40 people have never counted. They know who the experts are. They do not know how many tasks sit with exactly one person.

This post walks through the count for a hypothetical firm. The numbers are illustrative, not research, and the method takes about a week.

Why a firm of 40 is the awkward size

A firm of five carries everything in conversation. Everybody knows who does the payroll check because it is a five person room. A firm of 400 has process managers, written procedures and a continuity plan somebody owns.

A firm of 40 sits in between. It has outgrown shared memory and not yet built the systems that replace it. Work has been handed from partner to senior to preparer over the years, and each handover left a little know-how behind with the person who used to do it. Nobody decided that. It accumulated.

The result is a firm that looks healthy on a capacity chart and is fragile underneath. We covered the month end version of this in what happens to your close if one person quits. Here we look at the whole firm.

Mapping who can do what

Take a hypothetical firm of 40 with, say, 120 distinct recurring tasks across bookkeeping, payroll, VAT, management accounts and client reporting. If you have no list yet, build the recurring task list in a week first. The map needs it.

Then build a simple grid. Tasks down the side, people across the top. For each task, mark three levels:

1. Can do it alone. They have done it recently and would not need to ask anyone.
2. Could do it with help. They have seen it done, or done it once, and would need the manual or a colleague.
3. Does not know it exists. Leave blank.

Be strict about the first level. "Could probably work it out" is the second level. The test is simple: if the usual person were unreachable for three weeks, who would you hand it to without worrying?

Ask the people on the team to mark their own column and let them see the total. Self-marking takes less time than interviews, and it catches things a partner would never guess, such as the junior who quietly ran the supplier statement check last year.

Reading the count

Add up each row. For every task you now have a number: how many people can do it alone. Group the tasks by that number.

Imagine the hypothetical firm comes out like this:

• 22 tasks have four or more people who can do them alone.
• 38 tasks have two or three.
• 35 tasks have exactly one.
• 25 tasks have nobody at the first level. They are done by whoever is nearest, with help.

Those last two groups are the real answer. In this made up example, 60 of 120 tasks, half the firm's recurring work, depend on one person or on improvisation. That is the bus factor: for half the work, it is one.

The shape matters more than the total. Look for three patterns:

• Clusters on one person. If 14 of the 35 single-owner tasks belong to the same senior, you do not have 35 small risks. You have one large one, and a holiday you cannot afford to approve.
• Clusters on a client. Tasks that only one person has ever done for a particular client usually mean the client relationship is in one head too.
• Tasks nobody owns. The blank group is often the oldest work: the quarterly return nobody has touched since the person who understood it left.

What the count does not mean

A single-owner task is not a performance problem, and the person who owns it has not done anything wrong. They were good at it, so it kept coming to them. Often they are the most conscientious people in the firm.

Say that out loud when you share the map. If the first thing the team hears is "we found 35 risks", they will read it as an audit of themselves. If they hear "we found 35 tasks that are too much to carry alone, and we want to take some of that weight off you", they will help, and they will be right, because it is true. The aim is to spread know-how, not to make anyone easier to replace.

Closing the gaps, in order

Do not try to fix everything. Rank the single-owner tasks by what happens if they stop: a missed filing date, an unhappy client or a delayed invoice run. Start at the top.

For each one, the work is the same three steps:

1. Write down how the task is really done, with the person who does it, while they do it. Our post on stale SOPs explains why a manual written during the work beats one typed once and forgotten.
2. Have a second person do the task once with that manual, with the owner nearby.
3. Update the grid. The row moves from one to two.

There is a quiet benefit beyond continuity. A task with a written method and a second person who has run it is also a task you can discuss sensibly in terms of automation later. You cannot hand to an agent what you cannot describe, and a task that lives in one head cannot be described by anyone else. Documented work is what agents run on, and it is what a new colleague runs on too. This is the part of the job Beaks is being built to make routine, by keeping the map of tasks, owners and manuals in one place. But a spreadsheet will do for a first pass.

One thing to do this week

Pick ten recurring tasks you suspect are fragile. Write the names of the people who could do each one alone, honestly, without helping. Count how many have exactly one name. Then pick the one with the worst consequence if it stopped, and book an hour for its owner to show a colleague how it is done. One row moved from one to two is the whole method, repeated.