When not to automate: we'll tell you ourselves

This post will cost us projects. Good.

We're an automation consultancy, and we're going to tell you when automation is the wrong answer. Not as a trick — there's no "but actually you should" coming at the end. Sometimes the right answer is don't, and we'd rather you hear it from us before you pay anyone than figure it out after. This is the honest close to What automation actually costs (and what it pays back).

1. The volume is too low

Automation has a fixed cost and the problem has a monthly cost. If the monthly cost is tiny, the math never clears.

You do 12 invoices a month. Each takes 10 minutes. That's 2 hours a month — at $36/hr loaded, $72/month. A $4,000 integration pays back in 55 months. The tool you built it on will have changed twice by then.

Rule of thumb: if the payback period (here's the math) runs past 24 months, don't build it. Hire the hours, buy the cheap tool, or just keep doing it by hand. "By hand" is a legitimate strategy at low volume. Nobody says this because nobody gets paid when you hear it.

2. The process is broken

This is the big one. Automating a broken process doesn't fix it. It just makes the mess run faster.

If nobody agrees on how a job gets dispatched — if the morning dispatcher does it one way, the afternoon dispatcher does it another, and the owner does it a third way when he's covering — there is nothing to automate. You'd be encoding three conflicting processes into one system and then paying us to untangle the fights it causes.

> dispatch:??? →??? →??? ✗ nobody agrees

Fix the process first. Write down how it should work. Run it by hand for a month until everyone does it the same way. Then automate the version that works. The month of discipline is cheaper than the rebuild, and the automation will be simpler because the process is.

A consultant who offers to automate your process before it's consistent is selling you a very expensive lesson.

3. There's no owner

Every automated system needs one person who can answer the question: "It did something unexpected — is that right?"

Not a committee. Not "whoever's around." One named human who understands what the system is supposed to do and has the authority to say "yes, keep going" or "no, stop."

Without that person, here's what happens: the system hits its first real exception — the invoice with the weird line item, the booking that changed three times — and nobody decides. It sits. Or worse, somebody who doesn't understand it clicks "approve" on everything, and you find out in month three that it's been sending quotes to the wrong customers.

If you can't name the owner, you're not ready. Name them first. It's usually the person who already does the job by hand — they know where the bodies are buried.

4. Human time is genuinely cheaper

Sometimes the math just says no, and you should listen.

The task takes 30 minutes a month. Your admin costs $25/hr loaded. That's $12.50/month, $150/year. No automation project in the world beats that on price, and anyone who tries to sell you one is selling, not advising.

There's a subtler version too: tasks where the human touch is the product. The thank-you call after a big job. The personal note in the quote for a customer you've worked with for six years. Automating those doesn't save money — it spends trust. Your customers can tell.

What to do instead

"Don't automate" isn't "do nothing." For each case above, there's a right move:

• Low volume → keep doing it by hand, or buy the cheapest tool that handles it. Revisit when volume triples.
• Broken process → fix it on paper, run it manually for a month, then automate the working version. We'll still be here.
• No owner → name one. Give them authority over the system. Then build.
• Cheaper human time → let the human do it. Spend the automation budget where the hours actually hurt.

And sometimes the answer is "not yet." That's a complete answer. A good consultant will tell you what would have to change for the math to work — "come back when you're doing 40 of these a month" — instead of stretching the numbers to fit today's sale.

Why we publish this

Two reasons. First, it's true, and we'd rather be the consultancy that says it. Second, it's good business: every project we talk someone out of is a project that would have failed, and failed projects are what make small businesses distrust consultants. We'd rather have your trust for the project that actually pays back.

That's not a hiring problem. It's a wiring problem — and some wiring problems are solved by leaving the wires alone.

Get a free tool-stack audit

Bring us the thing you're thinking about automating. We'll run the math, check the process, and tell you straight — build it, buy it, fix the process first, or don't touch it. Book a free tool-stack audit.