How to Spot the One Process Actually Worth Automating First

Here's the uncomfortable truth nobody selling software will tell you: most of the tasks you'd love to automate aren't worth automating. They're annoying, sure. They eat your evenings. But annoying and worth-automating are two different things, and confusing them is the single most expensive mistake a small business makes when it finally decides to get organised. The skill that actually matters isn't picking a tool — it's learning to look at your own week and tell the real candidates apart from the decoys.

I've watched dozens of owners get this wrong in the same way. They pick the task that feels worst — the one that ruins their Sunday — and pour money into automating it, only to discover six weeks later that it happened twice a month and the automation paid for itself sometime around the year 2031. Meanwhile the dull, invisible task that quietly bled an hour a day went untouched because nobody noticed it. Feelings are a terrible guide here. Frequency, predictability and the cost of getting it wrong are good guides.

So this article isn't about how to automate. It's the step before that — the part everyone skips. How do you look at the messy reality of your business and reliably point at the one process that deserves to go first? Get this right and the rest is comparatively easy. Get it wrong and the best tool in the world won't save you.

Why the task that annoys you most is rarely the right one

Pain and value don't line up. The task that makes you groan loudest is usually the one that's emotionally heavy, not the one that's actually costing you the most. Firing a difficult email, dealing with a complaint, redoing a quote a client changed their mind on — these sting. But they're often rare, or they need exactly the kind of human judgement that automation can't replace. You remember them because they hurt, not because they're frequent.

The genuinely expensive tasks are almost always the ones you've stopped noticing. Copying a delivery address from an email into your order system. Re-keying the same supplier details every Monday. Checking whether three things matched before you send an invoice. None of these ruin your day. Each one takes ninety seconds. And you do them forty times a week without ever registering them as work. That's where the money is — in the stuff that's become invisible precisely because it's so routine.

This is why I never let an owner start from memory. Memory is biased toward drama. If you ask someone what wastes their time, they'll name the thing that frustrated them most recently — which is almost never the thing that frustrates them most often. You have to get the invisible tasks back into view before you can judge them, and that takes a little deliberate effort.

First, make the invisible work visible

You can't evaluate what you can't see, so before any scoring or judgement, you need a rough inventory of where your hours actually go. This doesn't need to be a formal time-and-motion study with a stopwatch. It needs to be honest and it needs to cover a normal week, not your idea of a normal week.

The cheapest method that works: keep a single open note for five working days, and every time you do something repetitive, add a tally mark next to it. New mark for a new type of task, a stroke for each repeat. Don't judge while you collect — just gather. By Friday the page tells you something your memory never would: which tasks repeat the most, and how the small invisible ones quietly out-total the dramatic ones you'd have named off the top of your head.

The five signals of a process worth automating

Once you can see your tasks, you need a way to judge them that doesn't rely on gut feeling. Over the years I've boiled it down to five signals. A process worth automating usually has most of them. A process that only has one or two is a trap dressed up as an opportunity.

1. It happens often

Frequency is the multiplier that makes everything else matter. A task that takes two minutes but happens fifty times a week is a far better candidate than one that takes an hour but happens once a month. Automation has a fixed setup cost, and frequency is what pays that cost back. If something happens only a handful of times a year, it almost never earns its place — no matter how much you dread it.

2. It follows the same steps every time

Automation loves a recipe. If a task runs the same way every time — same inputs, same steps, same output — it's a strong candidate. If it forks into a dozen special cases that each need a decision, you'll spend more time describing the exceptions than you'll ever save. The more sentences it takes to explain the task to a new hire, the worse a candidate it usually is.

3. It runs on rules, not judgement

There's a clean line between tasks that follow rules and tasks that require a human to weigh things up. Moving data between two systems is a rule. Deciding whether to give a loyal customer a discount is judgement. The rule-based tasks are the ones a machine can do reliably and tirelessly. Judgement-heavy tasks can sometimes be assisted by software, but trying to fully automate them early is how you generate confident, fast mistakes.

4. The inputs are already in a usable form

A task is much easier to automate when the information it needs already lives somewhere structured — a form, a system, a consistent email format. If the input is a pile of handwritten notes, voicemails or photos of receipts, automation is still possible, but you've quietly moved into harder, more expensive territory. Not impossible. Just not your first project. Save the messy-input tasks for after you've banked an easy win.

5. A mistake is cheap and easy to undo

Your first automation should be somewhere a small error is annoying, not catastrophic. A reminder that sends twice is embarrassing; a payroll run that fires twice is a crisis. Start where the blast radius is small. You want the freedom to let the automation run, watch it, and fix it in the open — not a process so critical that one bug ends up on your bank statement.

Turn the signals into a quick score

Signals are useful, but when you're staring at eight candidates you need something more decisive than a vibe. So turn them into a number. You don't need a formula a finance team would approve of — you need something fast enough that you'll actually use it. Here's the version I run with owners on the back of a napkin.

1. 1
   List your top candidates

   Pull the 5–8 tasks with the most tally marks from your week. Ignore everything below that — a long list just stalls the decision.
2. 2
   Score time-saved, 1 to 5

   How many hours a week would automating this genuinely give back? Be honest; multiply minutes-per-run by how often it runs, don't trust the dread.
3. 3
   Score how clean it is, 1 to 5

   Combine predictability, rule-based-ness and tidy inputs into one 'how automatable' number. A messy, judgement-heavy task scores low here even if it's frequent.
4. 4
   Multiply, then sanity-check the stakes

   Time × clean gives you a ranking. Then strike out anything where a mistake would be expensive or hard to undo — those aren't first projects, whatever they scored.
5. 5
   Pick the highest score you can finish soon

   Not the absolute top score — the top one you can realistically get live in two to three weeks. A finished small win beats an unfinished big one every time.

The multiplication matters more than it looks. A task that scores a 5 for time but a 1 for cleanliness lands at 5 — the same as a tidy little task that saves modest time. That's the method protecting you from yourself: it stops you chasing a big, painful, messy job just because the time-saved number looked exciting. The boring, clean, frequent task usually wins, and it usually should.

Warning signs you've picked the wrong process

Even with a score in front of you, it's worth knowing the smells of a bad candidate, because they're seductive and they all sound reasonable in the moment. If your chosen process trips any of these, stop and look again before you commit money to it.

• It takes more than a couple of sentences to explain. Complexity in the description means complexity in the build — and a lot of edge cases waiting to bite.
• You keep saying 'well, except when…'. Every exception is a branch someone has to handle, and exceptions multiply faster than you expect.
• It only happens a few times a year. However painful, the maths almost never works; you'll spend more setting it up than you'll ever save.
• It needs a human to feel something — empathy, taste, negotiation. Those aren't inefficiencies to remove, they're the job.
• A mistake would be expensive or hard to reverse. Great automation candidate, terrible first one. Earn your way up to it.

There's one more subtle red flag worth naming: automating a process you haven't actually agreed on. If two people in your business do the 'same' task three different ways, you don't have a process yet — you have three habits. Automating that just bakes the disagreement into software. Sort out how it should work on paper first; then, and only then, is it a candidate.

Spotting it isn't the same as it being ready

Here's a distinction that saves a lot of grief: a process can be the right one to automate and still not be ready to automate. Spotting the candidate is step one. Making sure it's in a fit state to hand to a machine is step two, and skipping it is how you end up automating a mess and getting a faster mess back.

A process is ready when you can write down exactly what goes in, exactly what comes out, and the steps in between with no 'it depends'. If you can't write those three things clearly, the task isn't a software problem yet — it's a clarity problem, and no tool fixes that. The good news is that the act of writing it down usually creates the clarity. Half the value of getting ready to automate is the thinking it forces you to do.

This is also the cheapest possible insurance against overspending. An afternoon spent describing a process on paper will tell you, for free, whether it's the clean ninety-second win you hoped for or a tangled judgement-call in disguise. Either answer is worth having before anyone quotes you for a build.

Where the good candidates usually hide

If you'd rather not wait a week to find out, there are a few places where strong first candidates show up again and again across almost every kind of small business. Treat these as a head start, not a substitute for looking at your own reality.

The seams between two systems that don't talk are the richest hunting ground — anywhere a human is the copy-paste cable between an email and a spreadsheet, or an order and an invoice. Close behind are the scheduled nudges: reminders, follow-ups, alerts, anything that's really just 'do this thing at this time, every time'. Then the repeated lookups — checking the same status, answering the same five customer questions, pulling the same Monday report from the same three places. None of these are glamorous. All of them are frequent, predictable and forgiving, which is exactly why they make such good first projects.

Common questions