· Sam Fielding · pricing · quotes · AI automation

How To Sanity-Check An AI Automation Quote

How to sanity-check an AI automation quote: the questions to ask, the red flags, the ownership test, and how to tell a real ops build from a dressed-up chatbot.

How To Sanity-Check An AI Automation Quote

You have a quote in front of you and no idea if it is fair. That is the normal position to be in, because most automation quotes are written to look impressive rather than to be checked. The fastest way to tell a good one from a bad one is not the dollar figure. It is whether the scope is named, the payback is shown, and you actually own what gets built. Run a quote through the questions below and you will know in ten minutes whether to sign, push back, or walk.

This is a practical checklist for a quote you have already received. If you want the ranges first, start with what AI automation costs in Australia. Here we assume you have the number and need to judge it.

The Bottom Line

  • The killer test for any quote: if you stop paying tomorrow, what do you keep? If the answer is nothing, you are renting, not owning.
  • A fair quote names the scope and deliverable, shows the payback maths, and lives on accounts in your name.
  • A single automation should land in the low thousands of AUD; a full install runs into five figures, with a small monthly run cost.
  • Red flags: no breakdown, lock-in, an open-ended retainer with no deliverable, and a chatbot sold as an operations system.

Start With What You’d Keep If You Stopped Paying

Ask one question before anything else: if I stop paying tomorrow, what do I keep? A good build lives on accounts in your name, so the Make.com and n8n scenarios, the logic, and the setup all stay with you. If the honest answer is “nothing”, you are renting a tool, not buying a system, and the rest of the quote barely matters.

This single test cuts through more bad quotes than any other. Lock-in hides in friendly language: “our platform”, “managed for you”, “hosted on our side”. Push the awkward version early. Ask who owns the accounts, where the build runs, and what happens to it the day the relationship ends. A straight answer means you are dealing with a builder. A vague one means you are about to rent something you thought you were buying. Walk through ownership properly in how agencies price automation.

Is The Scope And Deliverable Named?

A fair quote names what gets built and what it does. “AI automation system” is not a deliverable. “An Inbox Agent that triages your email and drafts replies for approval” is. If you cannot point at the finished thing and describe its job in one sentence, the scope is not real yet, and you are being asked to pay for a guess.

Look for a clear finish line on every line item. What triggers it? What does it touch? When is it done? A real proposal lists the automations, the tools each one connects, and where a human stays in the loop. A vague one lists outcomes and adjectives. The gap between “we will improve your operations” and “we will build a nightly routine that pulls your numbers into one place” is the gap between a bill you can check and one you cannot.

Watch for fixed prices quoted before anyone has seen your stack. Nobody can scope a build accurately without looking at your tools, your data, and your edge cases first. A flat rate handed over on the first call is a number pulled from the air, and you will pay for the guess one way or another. The things that actually move the price are covered in what drives the price of a build.

Does The Payback Case Stack Up?

A serious quote shows its working. The payback maths is simple: hours saved per run times your loaded hourly rate times runs per year, minus build and run costs. If a builder will not sit down and do that with you, that tells you something. They are selling a feeling, not a return, and feelings do not show up in your accounts.

Run it yourself on the back of a napkin. Say a task eats five hours a week and the person doing it carries a loaded cost of 60 AUD an hour. That is 300 AUD a week, roughly 15,000 AUD a year, sunk into one repeating job. If automating it costs a few thousand up front and a small amount monthly, it clears its own cost inside a quarter and keeps paying after. Use your real numbers, not these.

Be honest about the assumptions. If the maths only works when everything goes perfectly and every saved hour turns straight into revenue, treat that as a warning, not a green light. A fair quote uses a loaded rate, counts the run cost, and still comes out ahead with room to spare. If the return is thin even on paper, the build is not worth doing yet, whatever the demo looked like.

Run Cost Vs Build Cost

Check the relationship between the one-off build cost and the monthly run cost. Ongoing costs should stay small: tool subscriptions for Make.com and n8n, your AI usage, and light upkeep, together a modest monthly figure. The rule of thumb is blunt. If a “maintenance” quote rivals the build cost every single month, something is off and you should ask exactly what you are paying for.

A healthy quote separates the two clearly. Build cost is the work to design, wire, and test the automation. Run cost keeps it alive once it is shipped. When those two blur into one open-ended monthly number with no breakdown, you have lost the ability to check either. Ask for them split out. A builder who owns their pricing will hand it over without flinching.

The danger is the open-ended retainer dressed as run cost. “Ongoing optimisation” at a heavy monthly rate, with nothing shipped each month, is a subscription to hope. Run cost should be small and boring. If yours is large and vague, you are paying for a relationship, not a system.

Red Flags To Walk Away From

Some signals mean walk, not negotiate. No breakdown is the first: a single number with no line items hides whatever the seller does not want you to see. Lock-in is the second: a build that vanishes the day you stop paying was never yours. Together these two account for most of the five-figure regrets we hear about from operators who got burned before they found us.

Watch for the demo that dazzles and the ownership that never comes. A slick screen-share is easy to produce and tells you nothing about who owns the result. Watch for the open-ended retainer with no named deliverable, the “we will optimise things” with no list of what ships. And watch for a chatbot sold as operations. A scripted Q&A widget on your website is not a Daily Brief, an Inbox Agent, or a Command Centre. Paying chatbot money for a chatbot is fine. Paying operations money for one is not.

Here is your ten-minute checklist. Run every quote through it:

  • Ownership. If I stop paying tomorrow, do I keep the build? It must live on accounts in my name.
  • Scope. Is every deliverable named in one plain sentence with a clear finish line?
  • Payback. Does the quote show the maths, using a loaded rate and the real run cost?
  • Cost split. Is build cost separated from run cost, with run cost staying small?
  • Honesty. Did they price it after seeing my stack, or guess before they looked?

If a quote fails the ownership test, stop there. The other four do not matter if you do not get to keep what you paid for. If you are still weighing whether to buy a build at all, build vs hire vs DIY covers the decision before the quote.

Frequently Asked Questions

What Is The Single Most Important Question To Ask?

If I stop paying tomorrow, what do I keep? Ownership is the test everything else hangs off. A proper build lives on accounts in your name, so the scenarios, logic, and setup stay running whatever happens to the relationship. If the answer is “nothing”, you are renting a tool, and no amount of clever demo changes that.

How Do I Know If A Quote Is Too Expensive?

Compare it against the payback, not against another quote. Run the maths: hours saved times loaded rate times runs per year, minus build and run costs. A single automation should sit in the low thousands of AUD, a full install into five figures. If the build clears its own cost inside a year with room to spare, the price is fair.

Is A Fixed Price Before A Call A Good Or Bad Sign?

Bad, usually. Nobody can scope a build accurately without seeing your tools, your data, and your edge cases first. A flat rate quoted before anyone looks at your stack is a guess, and you pay for the guess later in scope creep or a thin result. A fair builder looks first, then prices what they actually saw.

Should The Monthly Cost Ever Match The Build Cost?

No. Run cost covers tool subscriptions for Make.com and n8n, your AI usage, and light upkeep, so it should sit well below the build figure. A maintenance retainer that rivals the build cost every month, or that names no deliverable in return, is a red flag. You are paying for things shipped, not for an open-ended relationship.

If you have a quote in front of you and want a straight read on whether it stacks up, map it against the checklist above and look hardest at ownership. We are happy to sanity-check the numbers with you and tell you plainly whether a build pays back, no homework, no surprise bill, no lock-in. Get In Touch.

Sam Fielding
Sam Fielding
Managing Director, Echelon AI Solutions

Sam co-founded Echelon AI Solutions and leads transformation strategy, client engagements and growth. He has built and operated businesses across marketing and AI education, and has guided companies in retail, trades, hospitality and professional services through operational change. His focus is making AI earn its place through measurable business performance.