AI Quoted Your Customer the Wrong Price

AI built the quote, the customer got a number, and it was wrong.

Now you have two bad options. Honor it and eat the margin, or walk it back and eat the trust. The tool didn't fail. It did exactly what it does: it produced a confident, well-formatted number from whatever rules it could find. And your pricing rules didn't exist anywhere but your head.

I run a door company in Dallas-Fort Worth, and I run Hivelead, where I help owners of $1M to $5M businesses figure out why they stopped growing. This is part of a series on when a business should actually use AI, and it's about the most expensive place to get it wrong.

How it happens

You were writing every quote yourself, from memory, and it was eating your evenings. So you pointed AI at it. Feed it the photos, the notes, maybe last year's invoices, and it builds the quote. Clean layout, line items, a total. It looks better than the ones you wrote by hand.

Then one goes out at a number that doesn't cover the job. Or a number so high the customer laughs and calls the next guy. Or the same job priced two different ways for two different customers in the same week.

Why the number was wrong

AI can only price from rules that exist somewhere. If your pricing lives in your head, the tool has nothing to work from, so it guesses. It averages old invoices, or it fills in a plausible figure, or it applies a rule you never meant.

And it doesn't know the things you know without thinking. That this customer is a property manager who's going to send six more buildings, so the first job is priced to win. That this door is on the second floor and the drive is forty minutes. That this homeowner is a price shopper who's going to call three more people anyway. You price all of that in your head, in about four seconds, every time. None of it was written down, so none of it made the quote.

Your Cost Per Lead Isn't the Problem. Your Average Job Is. covers why the ticket is decided by who you're quoting. AI doesn't know who it's quoting.

Not sure whether your leak is in the quote, the close, or somewhere upstream? The Revenue Leak Calculator puts a dollar figure on each of the five numbers. It's free and there's nothing to sign up for.

Honor it or walk it back

Once the number is in front of the customer, both options cost you.

Honor it, and you work the job at a margin you didn't choose. Do that a few times a month and it's real money, and it's invisible, because the jobs still show up as revenue. What's a Good Cost Per Lead? shows what a thin-margin job does to what you can afford to spend on the next lead.

Walk it back, and you've just told the customer the number they got from you can't be trusted. In a trade, that's the whole relationship. They may take the corrected quote. They won't refer you.

This is the amplifier problem at its most expensive

AI in the quote is AI in the conversation, and the quote is the one conversation where a mistake has a dollar figure on it. Faster follow-up into a dead quote wastes messages. A wrong price wastes the job.

The tool multiplied a process that was never written down, and what it multiplied was the gap.

The fix, in order

1. Write the price book. Not a spreadsheet of every possible job. The rules you actually use: base rate per job type, the adders (second floor, after hours, distance, commercial), the floor you won't go under, and the two or three customer types that change the number. If you can't write it down, the tool can't apply it. This is the same rule as every other piece in this series: fix the process by hand, then automate it.

2. Let AI do the arithmetic, not the judgment. Once the book exists, the tool is good at this. Photos and notes in, line items priced from your rules out. It handles the math and the formatting, which is the part that was eating your evenings. It does not decide who the customer is or whether to price to win.

3. A human reads every quote before it goes out. Thirty seconds. Does the number make sense for this customer and this job? That's the four-second judgment you were doing in your head, moved to the end of the process where it belongs. The tool drafts. You approve. Nothing goes to a customer that you haven't seen.

Do it in that order and the tool saves you the evenings and never costs you a job.

FAQ

Can AI write quotes for a service business?

It can do the arithmetic and the formatting once your pricing rules are written down: base rates, adders, the floor, and the customer types that change the number. It cannot make the judgment about who the customer is or whether to price a job to win. Write the price book first, let the tool draft from it, and have a person read every quote before it goes out.

What do I do if AI sent a customer a wrong price?

Decide quickly and once. If the margin is survivable and the customer is worth keeping, honor it and fix the process so it doesn't repeat. If it isn't, call the customer yourself, explain there was an error, and give the corrected number with something small to make it right. Either way, the fix is the same: get your pricing rules out of your head and into a document the tool can follow, and add a review step before sending.

Why did AI price the same job differently for two customers?

Because it had no rule to apply, so it guessed from whatever it could find, old invoices, averages, or a plausible number. Consistent pricing comes from a written price book, not from the tool. Once the rules exist, the tool applies them the same way every time.

Should the owner still review every quote?

Yes. The review is where the judgment you used to do in your head lives now: does this number make sense for this customer and this job? It takes about thirty seconds per quote, and it's the difference between AI saving you time and AI costing you a job.

Related reading

• Should You Use AI in Your Business?
• Stop Using AI to Talk to Your Customers

• Your Cost Per Lead Isn't the Problem. Your Average Job Is.
• What's a Good Cost Per Lead?