The wrong way to choose
When owners decide to automate something, they usually pick the task that annoys them personally. That's understandable and usually wrong. The most annoying task and the most valuable automation are rarely the same thing.
The better approach is to score your repetitive work like an investor, not a sufferer. I use five factors, each rated 1 to 5, and the arithmetic does the choosing. It takes about an hour, and it has saved my clients from expensive automations of things that barely mattered.
The five-factor score
Rate each candidate workflow on these five factors, 1 to 5:
- Frequency. How often does it happen? Daily scores 5; quarterly scores 1. Automation pays per repetition.
- Error cost. What does a mistake cost when a human does this tired on a Friday? A mistyped invoice amount or a lead sent to the wrong person scores high.
- Handoffs. How many people or systems does the work pass between? Every handoff is a place where things sit and wait. Three or more handoffs scores 5.
- Data quality. Does the workflow run on clean, structured data like form fields and spreadsheet columns, or on messy judgment calls? Clean data scores 5. This one is a gate, not just a score: automating messy data multiplies the mess.
- Reversibility. If the automation misfires, can you undo it? Sending a wrong internal notification is reversible; auto-charging a customer's card is not. Reversible scores 5, and your first automation should score at least 4 here.
Add them up. Anything scoring 20 or more out of 25 is a strong first candidate. Anything under 15, leave alone for now.
Scoring the usual suspects
Here's how the common candidates typically score for a service business, in my experience:
- Lead routing and first response: high frequency, high error cost since a lead that waits a day often buys elsewhere, clean data if leads come through a form, and completely reversible. Usually scores 21 to 24.
- Appointment scheduling and reminders: high frequency, moderate error cost, very clean data, reversible. Usually 19 to 22, and off-the-shelf tools handle it for 15 to 50 dollars a month.
- Invoicing: moderate frequency, high error cost, clean data, but reversibility is lower because money moves. Often 17 to 20. Automate the drafting and the reminders first, and keep a human on the send button.
- Reporting: often low frequency and low error cost. A monthly report that takes two hours scores maybe 12. Satisfying to automate, rarely the best first pick.
- Customer follow-up after a job, meaning review requests and repeat-business check-ins: high frequency, low risk, reversible. Usually 18 to 21, and it's the one that most directly grows revenue.
Why lead handling usually wins
For most service businesses I've looked at, lead handling wins the scoring, and there's a reason: it's the only workflow where minutes matter. A missed lead is invisible. Nothing breaks, no one complains, the revenue just quietly goes to whoever answered first. Automating the acknowledgment, the routing, and a next-day reminder turns your slowest response from days into minutes without anyone working harder.
One warning before you build anything: automation makes a process faster, including a bad process. If your current lead handling is whoever happens to check the inbox, fix the process on paper first: who owns which leads, what the first response says. Then automate the fixed version. Automating chaos gives you high-speed chaos.
What automation actually costs
Rough numbers, so you can budget:
- Off-the-shelf automation of one workflow, like a scheduling tool or a form connected to notifications: 0 to 100 dollars a month and a few hours of setup.
- A connector between two systems you already use, built on an automation platform: typically 500 to 3,000 dollars of setup work, then modest monthly fees.
- A small custom automation, for when off-the-shelf can't express your rules: usually 3,000 to 15,000 dollars.
Start at the top of that list and only move down when the cheaper tier genuinely can't do the job. The score tells you what to automate; the budget tiers tell you how.
What to do next
This week, do the exercise:
- List every task your team does more than once a week that follows the same steps each time. Aim for 8 to 12 items.
- Score each on frequency, error cost, handoffs, data quality, and reversibility, 1 to 5 each.
- Take the top scorer. Write down, on one page, exactly how it should work when it's done: who gets notified, what the customer sees, what happens when something fails.
- Spend one hour looking for an off-the-shelf tool that does that page. Only consider custom work if you can't find one.
- Run the automation alongside the manual process for two weeks before you trust it alone.
If you want a second set of eyes on your scoring, I do this exercise with DMV business owners regularly at HashWhales. It's a phone call, and the scoresheet is yours either way.