How to map a business process, and what not to automate

Search for “process mapping” and page one is mostly software companies explaining it: diagramming tools, project-management platforms and enterprise vendors, most of them selling the tool to draw it in. That makes sense for a large organisation with a team whose job is processes. In a small business the job is smaller and more practical. You want to see how one piece of work really gets done, find where it loses time or customers, and decide what to do about it.

This is how to do that in an afternoon, with no special software, and how to use the finished map to decide what is worth automating and what is not.

What is a business process map?

A business process map is a picture of how one piece of work moves from start to finish: every step, who does it, which tool it happens in, and where it waits. “From enquiry to booked job” is a process. So is “from finished job to paid invoice”, or “from new client to first delivery”.

A map is useful because most processes only exist in people’s heads. Each person knows their own part. Nobody sees the whole thing, so nobody sees the step where the same details are typed in twice, or the point where work sits for two days until someone notices it.

How to map a business process

1. Pick one process with a clear start and end. Name the trigger (“a customer fills in the form”) and the finish (“the job is booked”). Mapping “sales” or “operations” as a whole produces a diagram nobody reads.
2. Map how it runs today, not how it should run. This is the “as is” map. If a written procedure exists, it is rarely what people actually follow, and the gaps between the two are usually where the problems are.
3. Ask the person who does the work. They know the workarounds: the spreadsheet kept on the side, the reminder in their own calendar, the customer they always call instead of emailing. Workarounds are part of the process, so they go on the map.
4. Write down every step with three details: who does it, in which tool, and how the information gets there. Each time someone copies details from one place into another, mark it. That is where most mistakes start.
5. Mark the waits and the hand-offs. Anywhere the work passes to someone else, or sits until someone notices, note how long it usually waits.
6. Mark where it goes wrong. What gets missed, how often, and what it costs when it does. A missed enquiry is a lost customer. A late invoice is cash you do not have yet.
7. Only then draw the better version. This is the “to be” map: the same process with the problems fixed. Drawing it before the first map is how businesses end up automating a process that never worked.

Which tool should you use for process mapping?

For a small business, paper, a whiteboard or a spreadsheet is enough. A numbered list of steps with a name and a tool next to each one covers most processes. Diagramming software and formal notations such as BPMN earn their place when many people have to keep the same maps up to date for years, or when a process has so many branches that it no longer fits on a page.

Start simple. A clear map can always be redrawn in a tool later. A tool will not make an unclear map clear.

An example of a business process map

A window tinting company in Victoria runs its jobs in ServiceM8. Written out as a map, its enquiry process had two ways in:

• Website form: the customer fills in the contact form → someone copies the details into ServiceM8 by hand → a job is created.
• Phone: the customer calls → if someone is free, they take the details → if the team is out on a job or an install, the call rings out, and the customer may ring the next tinting company on their list.

Laid out like that, the two problems are plain: one step where details are typed in by hand, and one gap where customers are lost whenever the team is busy doing the work. Neither fix meant changing the software the business runs on. The contact form now connects directly to ServiceM8, so every enquiry creates a job with the customer’s details already filled in. When nobody is free to answer the phone, an AI agent takes the call and adds the caller’s details to ServiceM8 the same way. Two ways in, one place for every enquiry to land.

What not to automate

A finished map gives every step one of four answers: remove it, simplify it, automate it, or leave it with a person. Automation is only one of the four, and it is the wrong answer more often than the software vendors on page one would suggest.

Leave a step out of the automation, or deal with it another way, when:

• It should not exist. An approval nobody reads, a report nobody opens, a form that asks for details you already have. Automating a pointless step makes it faster and keeps it forever. Remove it instead.
• The process is still changing. If the way a job is done changes every month, an automation built today is out of date next month. Settle the process first.
• It needs judgement. Pricing an unusual job, handling a complaint, deciding whether a customer is worth chasing. These are the parts of the work people are paid for, and where a wrong automatic answer costs trust.
• It happens rarely. A task done twice a year usually costs more to automate and maintain than to do by hand.
• A mistake would be expensive and hard to spot. Payments, and anything with legal or safety consequences, need a person checking, even where parts of the work are automated.

The steps worth automating look different: repetitive, frequent, done the same way every time, and costly when they are missed. Details copied from one tool into another are the clearest case. So are reminders, confirmations and weekly reports put together by hand.

From the map to the first automation

Pick one step from the map: the one that is most repetitive and most costly when it goes wrong. Build it on the tools you already use wherever they can connect, rather than adding another app beside them. Test it on real cases before relying on it, and read what it actually does. Then go back to the map for the next step.

The same test works for AI. A practical AI strategy starts from the same list of jobs, and asks the same question of each one: is this worth handing over at all?

Where to start

You can map your first process this week with a notepad and the person who does the work. Pick the process that annoys people most, write it out as it really runs, and mark every re-typed detail, every wait and every place something gets missed. The map alone usually makes the first decision obvious.

If you would rather have help turning the map into working automation, mapping is the first stage of every business automation project at Doppio.