Most SMEs pick the wrong role to automate first. They go for the loud one, sales and marketing, when the real money is hiding in the boring one. Before you spend a dollar on AI, ask yourself these three questions.
We've onboarded 50+ Australian SMEs in the last 18 months. The single biggest predictor of whether AI saves them money in year one isn't the package they pick, it's the role they choose to automate first.
The owners who get this right see ROI inside three weeks. The ones who get it wrong spend $6,000 watching an AI cheerfully draft cold emails nobody reads.
Here's the test we walk every prospect through before our first call.
The answer is rarely the highest-paid person. It's usually the role that interrupts everyone else.
An admin on $60k who pings 4 people 20 times a day costs the business more than her salary suggests. Each ping is two minutes of context switch for the receiver, plus the recovery time. Multiply by your headcount and you'll find the real cost is closer to $180k of lost output.
The senior salesperson on $150k? She's expensive but she generates revenue. Don't automate her. Automate the role that's costing other people their day.
If the answer is "none of it, I'd hire someone first" â that's the role to automate.
You're already paying for it whether the seat is filled or not. Vacancy doesn't make the work go away. It just shifts it onto your operations manager, who is now drowning, who is now looking at recruiters.
An AI that handles 80% of that role at $300/month replaces the urgency of the hire and gives your ops manager their week back. That's a real cost saving, not a theoretical one.
Whatever stops working is the workflow worth automating. Everything else is noise.
If your bookkeeper takes a week off and invoices stop going out, you've found a workflow worth $4kâ8k/month of automation. If your customer service lead disconnects on a long weekend and your inbox triples, you've found another.
The test is simple: take a marker, walk through your last 6 months of operations, and circle every moment something quietly broke because one person was unavailable. Those are your automation candidates, in priority order.
A property manager we onboarded in February ran the test and was surprised. He thought he wanted AI for tenant communication, the loud, visible problem. The test pointed at maintenance dispatch: a workflow that touched 4 staff, broke every weekend, and cost him roughly 11 hours a week to manually coordinate.
We automated the dispatch in his first deployment. The system reads incoming maintenance requests, matches them to the right contractor by skill and availability, sends the quote back to the tenant, and only escalates exceptions. Day 7, it was running. Day 21, his weekly maintenance hours were down from 11 to 1.5.
The tenant chatbot? We built it in his second deployment, three months later, once the operational fire was already out.
If you don't have time to run the test, here's the heuristic that's right 80% of the time: automate the role nobody wants to talk about. Bookkeeping, dispatch, intake, scheduling, vendor coordination, compliance reporting. The unglamorous middle of your business is where the cheapest, fastest ROI lives.
Then come back to us when you want the customer-facing AI that gets all the attention on LinkedIn. By then you'll have the budget and the proof.
Most owners pick the AI workflow they'd brag about at a barbecue. Pick the one that breaks when someone's on leave. That's where the money is.