Framework 02

Process Map

Put a price on every process in your business, then rank the automations by how fast they pay for themselves.

The Problem

Every conversation about AI and automation starts the same way: this part feels slow, that part feels expensive, a vendor just pitched a tool for the other thing. The business ends up improving whatever is loudest.

The problem is that cost per run is not the same as cost to the business. A six minute task that runs three hundred times a month costs more than a two hour task that runs twice. A cheap process that fails one time in twenty may be your most expensive one once you count the cleanup. And the most expensive process on the list still isn’t automatically the one to automate, because the automation itself has a cost, and some processes are risky to hand to a machine.

Without the arithmetic, you’re prioritizing by feel.

The Idea

Every process that runs in your business has a monthly price: what it costs to run it, plus what it costs when it goes wrong. Price every process and the business stops being a feeling and starts becoming a map that tells you where your focus needs to be. You should be asking yourself:

• What does this process cost me every month?
• Can this be automated, and how much of it?
• What would the automation cost, and how quickly do I get my money back?

The Process Map is a table that answers those three questions for every process and turns each row into a verdict. Then you can sort the table by verdict, then by payback, and you have a clear roadmap for what to automate first, what needs a human in the loop, and what can be eliminated outright.

Here is what one looks like finished. One row per process, priced, qualified and ranked. Read it top down and it tells you what to build first, what needs a person watching it, and what to delete outright.

Three things fall out of it that no amount of discussion would have produced. Booking confirmations, not the loudest process, pay for themselves in under three months. Reconciling payments has a good payback but a yellow risk, so it ships with a human check no matter how good the number looks. And printing paper folios feeds nothing, so it is worth $292 a month for free, with nothing to build. The dashboard underneath turns the whole list into four numbers: what the business costs to run, what is recoverable, where the biggest leak is, and how fast the first win lands.

The rest of this module is how you fill that table in.

How It Works

Step 1: Price the Process

Every process has two costs:

• Run Cost: what it costs to actually do the job.
• Failure Cost: what it costs when the job is done incorrectly.

Add them together and you have the Monthly Process Cost: the full price of the process.

Run Cost takes three numbers:

• Frequency: how many times the process runs in a typical month.
• Speed: the hands on time for one run.
• Cost: what an hour of the operator’s time costs the business.

Multiply them, divide by 60 to put minutes and hours in the same units, and you have what you pay each month just to do the work.

Failure Cost takes two more:

• Error Rate: out of a hundred runs, how many come out wrong and need a second touch.
• Cost per Error: what one cleanup costs, the time to redo it plus anything the business eats, like a refund.

Multiply those by the frequency and you have the bill you’re already paying for things going wrong, usually without noticing.

Everything is priced per month, not per run. That’s how a six minute task can outrank a two hour one.

Step 2: Qualify for Automation

Not every expensive process should be automated. Three questions, no math, just judgment:

• Eliminate? Does anyone need the output? If not, delete the process instead of automating it. The cheapest automation is the one you never build.
• % Automatable: of one run, how much could a machine do end to end without a person touching it? This one takes some technical know how; if you don’t have it, ask someone who does.
• Automation Risk: how risky is it to let a machine run this? If an automated run goes wrong and nobody catches it, is that a nuisance, lost money, or real damage? Score it green, yellow or red.

Note that risk here is the risk of automating, not the difficulty of the task. A machine can check transactions better than any human; the question is what happens when it’s wrong and no one is watching.

Step 3: Rank for Automation

Ranking a process isn’t as simple as picking the most expensive one. You also have to consider the cost of building and maintaining the automation, and how quickly you see its returns. Payback captures all of it in one number: how many months until the automation has paid for itself.

Build Cost is what the automation costs to build, one time. It takes two numbers:

• Build Hours: how long the automation would take to build.
• Builder Rate: what an hour of the builder’s time costs, whether that’s you, an employee or a contractor.

Monthly Savings is what the automation earns back every month. It takes three numbers:

• % Automatable: the share of the process the machine takes over, from Step 2.
• Monthly Process Cost: the full price of the process, from Step 1.
• Ongoing Cost: what the automation costs to keep running, like tools, subscriptions and API fees.

Payback is Build Cost divided by Monthly Savings. The fewer the months, the sooner the automation pays for itself.

Step 4: Read the Map

That is the last number. Every process in the business now has a monthly cost, a monthly saving, a risk colour and a payback period, and every one of them was worked out the same way. Here is the Map at that point, in the order the processes were written down:

Eight rows, all of them true, and no idea what to do first. So how do you prioritize?

Go down the list and put every row through the same two questions. Is anyone using the output? If not, the process is eliminated, and you get that money back for free because there is nothing to build. Does automating it save anything? If not, skip it.

Whatever survives both questions is worth automating, and the risk colour from Step 2 decides how. Green gets the green light. Yellow and red are still worth doing, but they ship with a person watching, so they wait. Then, inside each colour, sort by payback, shortest first, so the money comes back as fast as it can.

Those are the five verdicts. Checked in order, top to bottom, same rules on every row, so two people looking at the same process reach the same answer.

Run the eight rows through the ladder and the Map sorts itself into a build order:

Get the Worksheet

The Process Map comes as two files. The worksheet is a seventeen page PDF: the steps above as instruction pages, with the Map redrawn at each checkpoint so you can see it fill in, then a page to score one process and a page to rank the whole Map. The calculator is a two tab spreadsheet that works in Excel or Google Sheets. Tab one is the Map: fill in the dashed blue cells, one row per process, and the grey cells calculate the costs, savings, payback and verdict. Tab two is the dashboard: your total monthly process cost, the recoverable savings, the biggest money leak and the fastest payback, with nothing to fill in. Both are free.

Working on something like this?

I build full stack AI systems for companies that want to improve how they work or do things that were not possible before. If you are working on this kind of problem, message me on LinkedIn or shoot me an email.

Where It Fits

The Process Map is the second of four tools. The Business System Blueprint makes the business visible: every process, mapped and scored. The Map prices those processes and ranks the automations, so you know what to automate first. The Map hands its top row to the Process Proposal, a one page pitch for the automation that gets a yes, a no, or a revisit-when, and the TRACE Method turns that yes into a build spec.

Each tool stands on its own. If you already have your list of processes, start here.