How to document a business process (without a six-month workshop)

Most companies document a process the same way. You book a room. You get forty people who touch the work into it. You put sticky notes on a wall for a week, then a few more weeks, then someone turns the photographs into a diagram.

It works. People do get to a map at the end. But it is slow, it is expensive, and the map starts going out of date the day it is finished.

Here is the part worth noticing: almost nobody in that room is inventing anything. They are remembering. And most of what they are remembering was written down years ago, by them, in a document that is still on the shared drive.

The three ways this actually gets done

There are really only three approaches.

The workshop. People in a room, working from memory. Sticky notes, a facilitator, weeks of sessions. You get one map.

The digital canvas. The same people, remote, still working from memory. Boxes on a screen instead of paper on a wall, with comments and version history. You still get one map.

Starting from the documents. The software reads what you already wrote, drafts the record, and a person confirms it. You get something you can ask questions of.

The first two look different and cost about the same. That is the whole point: moving the whiteboard onto a screen changed the surface. It did not change the input, and it did not change the output.

The first two ask people to remember what they already wrote down.

Why the input is the thing that matters

Think about what you are actually asking a room of people to do.

You are asking them to reconstruct, from memory, a thing that already exists in writing. The procedure document. The training pack. The screenshots someone made in 2019 to explain the awkward bit. The design document from the last system change.

It is all there. It is just spread across four hundred files, in a dozen formats, with no shared vocabulary and no way to search across it.

So the workshop is not really a mapping exercise. It is an archaeology exercise, staffed by the most expensive people in the building.

Change the input and everything downstream changes. Nobody is asked to remember. They are asked to confirm, and confirming is a much cheaper thing to do than remembering. It is also much easier to check, because every item shows the page it came from.

What the workshop is still for

This is not an argument that you should stop getting people in a room.

A room full of people does two things that no software can do. It surfaces the knowledge nobody ever wrote down: the exceptions, the workarounds, the reason a step exists that everyone has forgotten. And it gets people to agree on what they just saw, which is most of what makes the result stick.

That was always the valuable half. It was just buried under weeks of reconstructing the half that was already documented.

Do the eighty per cent that is written down in days, and spend the workshop on the twenty per cent that never was.

How to actually start

1. Find one process that matters and hurts. Not the easiest one. The one where the answers live in one person's head.
2. Gather what already exists for it. Do not clean it up. Do not reformat it. Awkward, inconsistent, half-finished documents are the normal case.
3. Get a draft out of them. The point is to start from something rather than nothing.
4. Have the people who know check it. This is where their time is worth spending. They are correcting, not reconstructing.
5. Only then book the room, for what is genuinely missing.

What you should end up with

Not a picture. A picture answers exactly one question: the one it was drawn to answer. Ask a different question and you commission another picture.

What you want is a record that answers the next question too. "Follow that invoice." "What does this role actually do?" "What changed after the policy update?" "Show me the controls." Same record, four answers, no rework.

And every part of it should point back at the page it came from, so anyone can challenge it, and being wrong is visible rather than buried.

See the three approaches side by side →

• Why process maps go out of date, and what to do about it

  The diagram was never the record. That one mistake explains why every map you have ever made started rotting the day it was finished.

• Process mining vs. process modeling vs. process context

  Three disciplines that get confused constantly. What each one actually tells you, what none of them tells you alone, and the order they belong in.