What Does Custom Business Automation Actually Cost? (Real Numbers, Published)

The question every seller dodges

“What’s this going to cost me?” is the second question everybody asks. The first is “have you done this before?” And the industry answers both the same way: with a discovery call, because a sales call is easier to close than a number.

I publish mine. Not ranges-that-hide, like “projects starting in the low five figures,” but actual numbers with the assumptions showing. Here’s the short version, then the honest long one.

The numbers

Custom business automation projects start at $7,000. That’s the smallest real build I’ll put my name on: one process, usually two systems, one direction.

Most projects land around $13,500. That’s the center of gravity: enough moving parts to be worth automating, not so many that scope runs wild.

Large multi-month builds have reached $35,000. That’s observed reality, not a menu: multi-system, bi-directional, compliance-heavy work that took a season. It’s also not a promise of scale; most problems never get near it.

Three sentences, and you now know more than most buyers learn in three meetings. The pricing page carries these same numbers as the canonical source; if a detail here and a detail there ever disagree, that page wins. It also carries the managed IT and hosting bands, if those are what you actually came for.

What $7,000 buys, and what it doesn’t

The entry build is one real process, end to end. A nightly sync that moves new invoices from the field system into QuickBooks, one direction, with errors flagged to a human. Or the weekly report pulled, assembled, and delivered before anyone clocks in. Small, but genuine: scheduled, error-handled, logged, documented, boring.

What $7,000 is not: everything, everywhere, both directions. Bi-directional sync across five systems with a compliance package on top is the $35,000 shape of the market, and anyone who quotes that work at $7,000 is planning to discover the difference later, on your invoice.

What the price is actually made of

A build price covers more than the code. Scoping: mapping the process precisely enough to quote a fixed number at all. The build itself. Error handling, which is where cheap quotes get cheap: the happy path is easy; the day the API hiccups is the part you’re paying for. Documentation, in plain English, so the thing isn’t a black box only its author understands. And handoff: your code, your systems, your documentation, no lock-in.

That last item is also why the market’s cheaper end exists. Marketplaces will quote a fraction of my numbers for “the same thing,” and once in a while it is. More often the difference is the parts you can’t see on a quote: what happens on failure, who answers at renewal, whether anyone documents a line of it. Price the whole lifecycle, not the invoice.

What moves the number

The full table lives on the pricing page; that’s canonical. The shape of it:

Drives cost up: more source systems (every additional one adds mapping and edge cases), dirty or legacy data (cleanup bills like work), compliance and encryption requirements, tight deadlines, formal service-level paperwork.

Drives cost down: clean data, one system-to-system hop, a flexible calendar, fast feedback while you QA alongside, and a reusable pattern: if I’ve built it before, you don’t pay for my learning curve.

The one-line version: half of these are yours to control before we ever talk. Hand me clean data, one hop, and a calendar instead of a fire drill, and the number drops before the first email.

Know your payback window before you commit

Price is half the equation. The other half is what the manual version already costs you — and most owners have never priced it. Don’t take my word for it; take ten minutes and run your own process through the calculator, pre-filled with a common example.

The pre-filled scenario: one person spending four hours a week on a manual process, at $28 an hour, weekly. That’s 208 hours and $5,824 a year in wages, before a single error, and errors are where these processes get expensive. Against that bill:

• The $7,000 entry build pays for itself in about 14 months.
• The $13,500 typical build, in about 28 months.
• Assume automation only hands back 60% of those hours (conservative, and my default assumption) and you’re at roughly two and four years instead.

Still usually the cheapest money in the building, but you should see both columns before signing anything. If your process is smaller than the example, every number above stretches; that’s not a pitch, it’s arithmetic.

The shape of this in the wild: a CFO whose Monday report ritual was 312 hours a year of one of the most expensive salaries in the building (case study), and a utility whose envelope routine ate 832 staff hours and $8,500 a year in stamps (case study). Your number is probably smaller than both — and still bigger than you think.

If $7,000 scares you

Then a custom build isn’t your answer yet, and that’s okay.

There are honest cheaper tiers. If the flow is small, one-directional, and low-stakes, middleware like Zapier or Make runs tens of dollars a month; I compared them against a dedicated script honestly here, including the times the cheap tool wins. And the do-nothing tier costs nothing new, because you’re already paying it, hidden in wages.

What I’d steer you away from is the middle move nobody admits to: buying a $7,000–$13,500 build while flinching, then starving it of scope until it’s an expensive half-solution. If the number scares you today, fix the process’s cost picture first: price the manual version, let the weekly pain get specific. Come back when it doesn’t. The number will still be published.

If you’re past the flinch and want your actual number, email me the process: what it eats in hours, which systems it touches. You’ll have a fixed quote, usually same day. That’s how the pricing page works too.