How to work out what to automate first: process map and priority matrix

There is a question that always comes up when you start automating: where do I begin? You have ten ideas, you do not know which pays off most, and the risk of spending…

2026-06-10 · Automation

There is a question that always comes up when you start automating: where do I begin? You have ten ideas, you do not know which pays off most, and the risk of spending weeks setting up something that does not matter is real. In this article I offer a simple method to identify the highest-impact automations for your specific business, before touching any tool.

The most common mistake: starting with the most interesting, not the most important

AI agents, complex flows across multiple systems, sophisticated integrations: they are technically fascinating and generate plenty of articles. But for most freelancers and small businesses, the biggest return is not there. It is in the mundane tasks that happen twenty times a day and that nobody finds interesting precisely because they are mechanical and predictable. Those are the ideal candidates.

The process map: the starting point

Before automating, you need to know what you are automating. That requires a map, even an informal one. Spend an hour answering these questions in writing:

Which are the five processes that consume the most of your time each week?
Which of those processes are always identical, with no significant exceptions?
Which depend on you or someone on your team remembering to do them?
In which are mistakes made with some frequency?

The answers to those questions already give you a list of candidates. Now you have to prioritise it.

The prioritisation matrix: impact vs. effort

For each candidate on your list, estimate two variables:

Impact: how much weekly time would automating this save? How many errors would it prevent? Does it improve the customer experience?
Effort: how complex is it to set the automation up? How many tools does it involve? Does it have complicated exceptions?

The logic is simple: start with the high impact, low effort automations. They are the ones that give you the most result for the least investment of time. The high impact, high effort ones come next. The low impact, high effort ones are probably not worth it.

The signs that a process is ready to be automated

Not every process that looks automatable is a good candidate right now. These signs indicate that a process is mature:

It always happens the same way: if you have to think each time to decide what to do, the process is not standardised and automating it is premature.
You can describe every step exactly: if someone asks you what happens from start to finish, you can answer without hesitating.
You have already done it enough times: new processes change. Automating something you are still adjusting is building on sand; wait for it to settle.
The value is in it happening, not in how it happens: if the process requires judgement or adaptation case by case, not yet.

An example: how to prioritise in a small consultancy

Imagine a two-person consultancy with these routine manual processes: answering queries by email, recording clients in a spreadsheet, sending monthly reports, arranging meetings and following up on proposals sent.

Applying the matrix: arranging meetings has high impact (lots of back-and-forth emails) and low effort (a scheduling tool set up in 20 minutes). It goes first. Recording clients in the spreadsheet has high impact (repeated data, frequent errors) and low effort (a form connected to the spreadsheet). It goes second. The monthly reports have high impact but medium effort (they need some customisation). They go third. And so on.

The most useful practical rule

If you do not know where to start, ask yourself a single question: what interrupts you most, or what annoys you most about having to do it by hand this week? The answer almost always points to the automation with the greatest perceived impact. Start there. It is not the most sophisticated method, but it is the one that most often leads to a useful first automation in the least time.

What changes once you have your first automation running

Something happens with the first successful automation that no tutorial and no article delivers: you start seeing your own business through a different lens. Instead of viewing tasks as something that "has to be done", you automatically start asking whether a system could do it. That change of perspective is worth as much as the hour you recovered.

After that first concrete win, the second automation is easier. You know how the process of mapping, prioritising and configuring works. It is no longer unknown territory. And the third is easier still. Over time, that habit of spotting what can be systematised becomes natural, and the business gains operating capacity without you having to hire anyone or increase your workload.

That is exactly what AutoFlow is after: that the first step is clear, that the second is easier than the first, and that the process sustains itself.

In short

To identify where to automate, first map your repetitive manual processes and then prioritise by impact versus effort. Start with the high impact / low effort quadrant. Only automate processes that are already standardised and that you can describe step by step. And if you need a shortcut: start with whatever interrupts you most or annoys you most to do by hand today.

— AutoFlow · Automation and AI for small businesses and freelancers.

If this sounds like you

Drawing this map with you and coming out with a prioritised list of what to automate, in what order and what each piece costs is a fixed-price piece of work: what you can automate with AI. What gets built afterwards is automation with AI.