There is an old line about automation: applied to an efficient operation it magnifies the efficiency, and applied to an inefficient one it magnifies the inefficiency. It has survived because it keeps being true.
The most expensive mistake in this field is paying to build a fast version of a process that should not exist in its current form. An hour of mapping prevents it. This is the method we use in our free audit, written out so you can run it yourself.
Get the right people in the room
The person who actually does the work, not their manager. Managers describe the process as designed; the person doing it describes the process as it is, including the workaround they invented in 2023 that everyone now depends on.
If several people do the same job, get two of them. The differences between their descriptions are the most valuable output of the whole exercise.
The five questions
1. What starts this?
The trigger. An email, a form, a date, a phone call, someone noticing something. Be specific — "when a customer orders" is often really "when the warehouse WhatsApp group gets a message".
2. What are the steps, in order?
One line each, with the system used and roughly how long it takes. Do not tidy as you go. You want the real sequence including the double entry and the spreadsheet nobody admits to.
3. Where does it wait?
Mark every point where work sits idle waiting for a person, an approval or an external party. In most processes waiting dwarfs working — and waiting is usually cheaper to fix.
4. What goes wrong, and how often?
The exceptions. Every process has them and they are usually where the real time goes. "About one in ten needs chasing" is the kind of detail that changes what you should build.
5. What happens to the output?
Follow it to the end. Reports that nobody reads are surprisingly common, and the correct automation for a report nobody reads is to stop producing it.
Now go looking for steps to delete
This is the part that pays for the exercise. Before automating anything, ask of each step:
- Why does this exist? If the answer is "we've always done it", investigate. Some steps outlive their reason by years.
- Who consumes this? If nobody, delete it. Free.
- Is this checking something that could be prevented? Verification steps often exist because an earlier step allows a mistake. Fix the earlier step and the check disappears.
- Is this data entered somewhere it already exists? That is an integration, not an automation.
- Would a simpler form remove this? Half of chasing is caused by requests that were unclear the first time.
Deleting a step costs nothing and has no running cost. It is strictly better than automating it. A good vendor will find these and tell you, even though it reduces the size of the build.
Then decide what to automate
What remains, ranked by hours saved per dollar spent. Good first candidates share a shape:
- High frequency — daily or weekly beats monthly.
- Rule-based, or judgement that can be described clearly.
- Structured or semi-structured input.
- A clear right answer, so you can measure whether it worked.
- Low blast radius if it goes wrong.
Poor first candidates: rare, highly variable, requiring real expertise, or where an error is expensive and hard to detect. These are not un-automatable — they are bad places to start.
What the map should look like when you are done
A single page. Trigger at the top, steps in order with system and duration, waiting points marked, exceptions listed with rough frequency, and the output at the bottom. Deleted steps struck through rather than removed, so the reasoning survives.
That page is what you hand a vendor. It will get you a more accurate quote than any discovery call, because it removes the guesswork they would otherwise price for.
Frequently asked questions
How long should mapping take?
An hour per process for most business workflows. If it is taking a day, the scope is bigger than one process and should be split.
Do I need special software?
No. A sheet of paper or a shared doc is fine. Flowchart tools add polish, not insight, and they tempt people to tidy the process while documenting it — which defeats the point.
What if nobody agrees on how the process works?
That is the finding, and it is a common one. Different people doing the same job differently means you have variation to resolve before automating — otherwise you are encoding one person's version as company policy by accident.
Will a vendor do this for us?
A good one will, and should include it. Ours is free and comes back with a ranked list including the honest answer if nothing is worth automating yet. Be wary of anyone who quotes before mapping — that number will move.
Want us to map it with you? The audit is free, takes about 30 minutes, and produces a ranked list by hours saved per dollar. See how it works or message us on WhatsApp.
