Insights
What should I automate first? Start with the leak, not the tech.
The right first automation is whatever's leaking the most money for the least effort to fix. Not the most impressive thing you could build — the leak.
Every business leaks in one of three ways: money you should be earning and aren't, money you're spending that you shouldn't be, and hours of your own time going to work someone — or something — else could handle. Run your last month through those three buckets and the candidates name themselves.
That's the method in two paragraphs. The rest of this piece is the working version: what belongs in each bucket, how to put rough numbers on them with an afternoon's counting, and how to rank what you find — because the ranking is where first-automation decisions actually go right or wrong.
Bucket one: what money should I be earning that I'm not?
This is revenue that reached for you and missed. Count four things from last month. Calls that rang out — after hours, during jobs, lunch rushes — and how many of those callers you ever heard from again (what catching them actually involves). Estimates that went quiet — sent, never followed up, never closed; this one deserves its own arithmetic, because for most service businesses it's the biggest single line. No-shows and unfilled slots — appointments that evaporated without a reminder, and the calendar holes they left. Past customers gone silent — the service-due, season-due, overdue-for-a-checkup list nobody works.
Rough pricing is enough: your average ticket, your normal close rate, count times rate times ticket. You're not building a court case; you're building a ranking — and a ranking built on rough real numbers beats a precise guess every time.
Bucket two: what am I spending that I shouldn't be?
Quieter, easier to miss. Software you pay for and don't use — including, commonly, tools whose features overlap so you're paying twice for one function. An hour with your card statement finds it. Rework from dropped handoffs — the job scheduled twice, the part ordered twice, the invoice that went out wrong because information got retyped between systems; most of this lives in the seams between your tools. Interest and late fees you're eating because your own invoices go out slow and get chased never. Paying premium rates for overflow — the answering service, the temp, the rush vendor — to handle volume your systems drop at predictable times.
Bucket three: where do my own hours go?
The most mispriced asset in a small business is the owner's evening. Count your week honestly: invoicing and bookkeeping-adjacent shuffling, retyping information from one system into another, answering the same customer questions for the hundredth time, chasing your own crew for job status, building the schedule. Price those hours at what an hour of yours is actually worth — not zero, which is the rate most owners quietly use. There's a whole piece on that arithmetic, but for ranking purposes: count of hours, times your rate, per month.
None of this is exciting, which is exactly why it's still broken. The boring leaks survive because nothing about them demands attention today — they just cost you a little, every day, forever.
How do I rank what I found?
Two axes: what it costs monthly (you just counted that) and how hard it is to fix. For effort, three honest questions per leak. Is the pattern stable? Chasing an estimate looks the same every time — cheap to automate. Handling a custom commercial negotiation doesn't — expensive, maybe never. Do the systems involved connect? A leak between two modern tools with open doors is a plumbing job; a leak inside a fossil system is a project. How bad is a mistake? A follow-up text that lands weird is survivable; an automated invoice that bills wrong isn't. Higher stakes mean more build care, more testing, more cost.
Now put the two axes together: big leak, easy fix — do it first, and there's usually one obvious winner staring at you from bucket one. Big leak, hard fix — second tier, worth real money, worth doing deliberately. Small leak, easy fix — fine as a warm-up if the winner scares you. Small leak, hard fix — leave it, and feel good about leaving it; declining to automate something is a decision, not a failure.
One more discipline, cheap and almost universally skipped: date the ranking and re-run it in a quarter. The first fix changes the list — plug the missed-call leak and the follow-up leak behind it suddenly shows its real size — and a business that re-counts a couple of times a year automates in the order its numbers actually move, instead of the order the demos arrived in.
And a caution from the other direction: the worst first automation is usually the most interesting one. The AI-sounding project — the chatbot that answers anything, the dashboard that knows everything — is the one that demos best, goes stale fastest, and touches no leak you counted. If a candidate didn't come out of your buckets, it's not a candidate. It's entertainment.
Notice the list still has no technology in it. What to automate first is arithmetic — which leak costs the most, which fix is cheapest to get right — crossed with something no tool supplies: knowing your business well enough to tell a real leak from a loud one. That ranking is the judgment call, and it's the part worth getting right before anyone builds anything.
The counting above is yours to run, and I mean that — an owner with an afternoon and a legal pad can get ninety percent of the way. The last ten percent — which leak, in what order, at what tradeoff for your shop specifically — is the call that pays, and it's the work I do. Bring me your three buckets: book a conversation, and I'll tell you straight what I'd fix first, and what I'd leave alone.