How to automate your business processes — the 7 steps that actually stick
Most automation fails because it starts with the tool instead of the process. The 7-step method I use with every client: find the repeat, put a number on the leak, fix the process, then automate the smallest piece first.

Daniel Huthwaite
Founder, 88 Automate
Most business owners who ask me about automation start with the wrong question. They ask "which tool should I use?" The right question is "which process should I fix first?" Pick the wrong process and even the best tool automates a mess. Pick the right one and almost any decent tool pays for itself in a month.
This is the method I use with every client, in the order I use it. It's the same whether you do it yourself or hire someone — the only difference is who does the thinking.
Step 1: Find the process where work repeats
Automation pays when a job happens often, the same way each time, and nobody enjoys doing it. The easiest place to look is wherever the same information gets retyped. Sales enquiries into a spreadsheet. Spreadsheet rows into invoices. Invoices into the accounts package. Each retyping is a chance to be late, wrong, or both.
Sit down with your team and ask one question: "what do you do this week that you also did last week?" Everything on that list is a candidate. You don't need process maps or special software for this — a shared list is enough.
Step 2: Put a number on what it costs you
Before automating anything, work out what the process costs today. Not in theory — in hours and money. If two people spend forty minutes a day retyping orders, that's over 300 hours a year. At £20 an hour, that's £6,000 of wages spent copying data. If late quotes lose you even one order a month, add what that order is worth.
This number does three things. It tells you which process to fix first (the biggest leak, not the easiest one). It tells you what a fix is worth, so you know what it's sensible to spend. And it gives you the benchmark to check the automation actually worked. I put the audit at the heart of my work for exactly this reason — a £2,000 fixed-price audit that finds these numbers before any software is chosen.
Step 3: Fix the process, then automate it
This is the step most people skip, and it's the one that decides whether the automation survives. Automating a broken process gets you the same mess faster. If three people each have their own way of logging a job, no tool fixes that — the tool just enforces one of the three ways without anyone agreeing to it.
So before you build anything, agree on the process itself. Who owns each step? What happens when something unusual arrives? What gets checked before it moves on? Write it down in a page. It'll be wrong in places; that's fine. Fixing a page is easier than fixing software.
Step 4: Choose the tool that fits the process, not the other way round
Only now look at software. A few honest rules:
- Start with what you already pay for. Most small businesses already own capable tools they barely use — CRM features sitting inside their accounts package, forms inside their email platform, AI inside the software they already have.
- Prefer one system that owns each piece of data. If your customer details live in three places, no automation will keep them in agreement. One system owns each fact; everything else reads from it.
- Be suspicious of anything that needs a new subscription per step. A chain of five specialist tools means five bills and four connections to break. Two good tools connected properly usually beat five specialist ones.
- Zoho is my core stack because for a business of 5 to 50 people it does what the big names do, the apps are built to work together, and the licences cost £20–£50 per person per month rather than triple that. But the principle matters more than the brand: fit the tool to the process, never the reverse.
Step 5: Build the smallest version first
Don't automate the whole process in one go. Automate the single most annoying step, run it for two weeks against the old way, and compare. Did it save the hours you predicted? Did anything break that nobody noticed?
The first automation I build for a client is almost always small — an enquiry that lands in the CRM and pings the right person, an invoice that issues itself when a job closes. Small automations earn trust. Big-bang projects that touch everything on day one are where people get burned, and where the team quietly goes back to their spreadsheets.
Step 6: Give every automation an owner and a check
An automation with no owner is a bomb with a long fuse. The tool changes its interface, a field gets renamed, a supplier changes their export format — and nobody notices until a customer does.
Every automation needs three things: someone whose job it is to watch it, a simple weekly check that the numbers it produces match reality, and a rule that new automations get documented. This is why every build I do includes monitoring — I've been on the wrong end of a checkout that refused payments for hours while the tool showed green. Problems find the watcher before they find your customers, or they don't get found at all.
Step 7: Then look again for the next one
Once the first automation is running and saving real hours, repeat from Step 1. The second process is easier to fix than the first, because the team has seen one work and the systems now talk to each other. That's the compounding part: each automation makes the next one cheaper.
Where AI fits in
AI is the newest layer, and it's worth being clear about what it's good at today: reading and writing text that would take a human minutes. Drafting follow-ups from CRM notes. Summarising a thread before it lands on your desk. Turning a jumbled enquiry email into structured data your system can route.
What it isn't, yet, is a decision-maker — and a voice agent or an auto-replying chatbot is a maturity level you earn, not a first step. Put AI on top of processes that already work. If the process is broken, AI just breaks it faster and more confidently.
The honest shortcuts
If you take nothing else from this: start with the process, not the tool. Put a number on the leak before you spend anything on fixing it. Fix how the work runs before you automate it. Give every automation a watcher.
That's the whole method. It's how I work with clients, and it's how you can start on your own this week — the first two steps need nothing but a list and a calculator. And if you want the diagnosis done for you, that's what the audit is for.
Sound like your business?
Thirty minutes on a call with me. You tell me what's frustrating you, I tell you straight what I'd do.
More notes
Five apps, and no screen showed the whole client
A bookkeeping practice with 84 clients and five good tools that didn't talk. Why we diagnosed before building, and why we didn't migrate a single record.
AI writes the code now. So what am I for?
Building software gets cheaper every six months. An honest look at what that does to a one-person automation consultancy, why cheap to write is not the same as safe to run, and where the value actually sits.