The AI mistakes that cost tradespeople jobs are: quotes with figures the AI made up, phone bots that don't recognise an emergency, instant replies that don't answer the customer's question, AI photos passed off as real work, fake or filtered reviews, follow-ups that turn into nagging, and website copy promising things you can't deliver. Each is easy to spot once you look.
What they have in common is that the AI did what it was asked, fluently, and nobody checked. Trades work runs on trust built quickly: a customer usually contacts two or three firms and picks the one that seems most competent and straight with them. AI can make you faster than the others. Used carelessly, it makes you look careless, and customers rarely tell you why they went elsewhere.
1. Letting the AI put numbers into your quotes
Ask a chat assistant to "write a quote for repainting a hallway, stairs and landing" and it will produce a tidy document with quantities, days and prices. None of those numbers came from you. It doesn't know your day rate, your paint supplier's prices or how many coats that stairwell needs.
How it loses the job: an illustrative decorator sends an AI-drafted quote estimating two days and 10 litres of paint. The customer's other quote says three days, lists the filling and sanding separately, and explains why. The customer trusts the detailed one. Or the opposite happens: the AI's quote is accepted, the job takes three days, and the decorator works the third for nothing. Missing lines are as common as wrong ones: waste disposal, access equipment, making good after an electrician's chasing.
The fix: numbers come from your price list and your measurements, every time. The AI's job is wording, layout and making sure nothing is missing from the list. A prompt line that works: "Use only the prices and quantities I give you. If something seems to be missing, list it as a question for me; don't add a price."
That last instruction earns its place. Given the decorator's own figures (three days at the day rate, paint and materials at cost, walls and woodwork in two coats), an illustrative reply came back with the quote laid out properly and three questions underneath:
Questions before this goes out:
1. The stairwell wall is about 5 m high at its tallest. Is access
equipment (tower or platform) included, or priced separately?
2. You mentioned cracks above the front door. Is filling included
in the three days?
3. Who lifts and relays the stair carpet, if it's being protected
rather than removed?Question 1 was the one that mattered: the decorator had forgotten the tower hire. Added as its own line, it made the quote more expensive and more convincing, because it showed someone had thought about how the stairwell would actually be reached.
What tradespeople should check before trusting an AI price goes through the checks line by line.
2. The phone bot that didn't recognise an emergency
AI receptionists and answering services are a genuinely good fit for trades, because you can't answer the phone from a ladder. The mistake is setting one up with only the happy path in mind.
How it loses the job: a caller says "water's coming through the kitchen ceiling". The bot, trained to book appointments, offers Thursday at 10am. The caller hangs up and rings the next plumber on the list, who answers. You don't even know the call happened unless you read the transcripts.
The fix: give the system a list of phrases that transfer straight to your mobile or trigger an urgent text (water coming through, no heating with a baby or elderly person in the house, smell of gas, sparking, no power, locked out, roof leaking in heavy rain). Then test it by calling yourself with each phrase. Read a week of transcripts every month.
That monthly read is where the list improves, because callers rarely use your phrases. An illustrative week for an electrician's answering service: 31 calls, 26 booked or answered correctly, two passed on as urgent, three booked for later. Of those three, two were fine (a quote for outdoor lights, a question about an invoice). The third said: "The kitchen lights keep flickering and there's a funny burning smell near the switch." The bot booked it for the following Wednesday because neither "sparking" nor "no power" was mentioned. Add "burning smell", "hot socket" and "scorch marks" to the list, ring the caller back that day, and test the new phrases with a call to yourself. Setting up an AI answering service that never misses an emergency has a full escalation list to copy.
3. Instant replies that don't answer the question
Speed wins trade enquiries, so plenty of firms switch on an AI auto-reply. The trap is replying fast with nothing useful.
Customer's message: "Can you fit an EV charger at a semi-detached house next week? Roughly what does it cost? Driveway's at the front, fuse box is in the hall."
Generic AI reply (loses the job): "Thank you for your enquiry! We pride ourselves on quality workmanship and excellent customer service. A member of our team will be in touch shortly."
Specific reply (wins the survey visit): "Thanks, yes, that's a straightforward setup with the fuse box in the hall. Most installs like yours come in between [your range], depending on cable run and whether the board needs work. We have Tuesday or Thursday next week. Could you send a photo of the fuse box so we can confirm?"
The first reply was instant and worthless; the customer is still waiting for an answer and has already messaged two other firms. The second answers all three questions, uses your real price range (supplied by you, not the AI), and asks for the one thing that moves the job forward. The difference is the instructions: give the AI your services, typical price ranges and availability, and tell it to answer the customer's actual questions before anything else. Replying to trade enquiries in 60 seconds with AI shows the setup.
4. AI photos passed off as your own work
AI image tools can produce a gleaming kitchen, a perfect patio or a "before and after" that never happened. Some firms use them to fill out a thin portfolio, or "enhance" real photos until they're unrecognisable.
How it loses the job: the customer arrives for a site visit, or you arrive at theirs, and they ask about the job in the photo. You can't say where it was. Or a customer runs a reverse image search and finds the same picture on three other firms' websites. Either way, your honesty is now in question on everything else you've said, including the price. Presenting images of work you didn't do as your own can also fall foul of advertising and consumer-protection rules in many places.
The fix: real photos of real jobs only in your portfolio, even if they're phone snaps. If you use an AI image to illustrate an idea ("what a porch like this could look like"), label it clearly as an illustration. A slightly messy real photo with a caption like "hallway, stairs and landing, three days, two coats" is more persuasive than any generated image.
The grey area is AI editing of real photos, and a simple line works: change the light, never the work. Brightening a dim bathroom, straightening a wonky angle or cropping out a skip is fine. Removing a visible defect, or letting an "enhance" button tidy the grout lines on a tiling job into perfect, even rows, isn't, because a customer will compare your photo with what they see on their own wall. A realistic way it surfaces: a tiler's gallery photo, run through a phone's AI enhance feature, shows grout joints noticeably thinner and straighter than the tiler actually lays them. A customer who books on the strength of it notices the difference on day one and starts questioning everything else.
5. Fake reviews and filtered review requests
AI makes it trivially easy to write twenty glowing reviews in twenty different voices. It also makes it easy to send review requests only to customers you think were happy. Both are against Google's rules: its policy for Maps content prohibits content not based on a real experience, and says businesses must not discourage negative reviews or selectively solicit positive ones. Reviews written by you, your staff or anyone with a business relationship also count as a conflict of interest. Buying or writing fake reviews is also increasingly illegal under consumer law in many markets.
How it loses the job: Google removes the suspicious reviews and can restrict your profile; a competitor or customer notices the pattern and says so publicly; or a customer who wasn't asked for a review, because the job went badly, finds another way to complain. The quieter version is AI review replies that are all identical ("Thank you so much for your kind words!"), which makes genuine reviews look manufactured too.
The fix: ask every customer, automatically, after every job, and let the reviews be what they are. Use AI to draft replies, but make each one mention something specific about the job. What review gating is and how it gets businesses penalised explains the rules in detail.
6. Follow-ups that turn into nagging
Following up on quotes is one of the best uses of AI in a trades business, because most tradespeople don't follow up at all. Automate it badly, though, and you swing from silence to harassment.
How it loses the job: an illustrative roofer sends a quote on Monday. The automation sends "Just checking in!" on Tuesday, Thursday, Saturday and the following Monday, each with the same wording. The customer, who was waiting for their partner to get back from a work trip before deciding, picks the other roofer out of irritation and mentions it on a local community group.
The fix: two or three follow-ups at most, spaced out (for example day 3, day 10 and day 21), each with a different, useful reason to reply: answering a common question, offering a date, or simply asking whether they'd like you to hold the price. Stop the sequence the moment the customer replies, even to say "not yet". For the roofer, a sequence that would have kept the job might read:
Day 3: "Hi [first name], a quick one on the roof quote: a few people ask whether the price includes the scaffold. It does, erection and removal. Happy to answer anything else."
Day 10: "Hi [first name], we've got a gap in the diary in the week of [date] if that suits. No pressure; just let me know if you'd like me to pencil you in."
Day 21: "Hi [first name], I'll hold the quoted price until the end of the month. If the timing's not right, no problem at all, and the quote's there when you need it."
Each message gives a reason to reply, none repeats the last, and the third closes the loop politely instead of leaving the customer to dread the fourth. A decorator's playbook for following up silent quotes has wording for each touch.
7. Web and profile copy that promises what you can't deliver
Ask an AI to write your website or business profile and it fills gaps with the standard claims of your trade: "24/7 emergency call-outs", "fully certified for all gas and electrical work", "no job too small", "lifetime guarantee", a list of services you don't offer and towns you don't cover.
How it loses the job, or worse: a customer rings at 2am because your site says 24/7, gets voicemail and leaves a one-star review. Someone books you for work you're not qualified or registered to do because the site said "all electrical work". A customer quotes the "lifetime guarantee" back at you three years later. Claiming a certification or membership you don't hold can have consequences well beyond one lost job.
The fix: before the AI writes anything, give it a fact sheet: services you do and don't offer, areas you cover, hours, your actual qualifications and memberships, your guarantee terms. Tell it to use only those facts. Then read every claim on the page and ask, "could I prove this if a customer challenged it?"
A filled-in fact sheet for an illustrative two-person electrical firm, short enough to paste into any prompt:
SERVICES WE DO: domestic rewires, consumer unit upgrades, EV
charger installs, fault finding, extra sockets and lighting.
SERVICES WE DON'T: commercial or three-phase work, alarms, CCTV.
HOURS: 8am-6pm Monday to Friday. Emergency call-outs for existing
customers only, until 9pm.
AREA: within about 25 miles of our base.
QUALIFICATIONS: [your registration or scheme, exactly as held].
GUARANTEE: 12 months on workmanship; manufacturers' warranties on
parts.
NEVER CLAIM: 24/7 service, "all electrical work", lifetime
guarantees, awards or memberships not listed here.The "never claim" block matters most, because it names the exact phrases the model reaches for when it's filling space.
What these mistakes cost an illustrative two-person electrical firm
It's easy to wave these off as small. Consider an illustrative two-person electrical firm that gets about 40 enquiries a month, wins about 30% of the jobs it quotes, and averages $650 a job.
- Generic instant replies (mistake 3) mean fewer enquiries turn into survey visits. If that drops the win rate from 30% to 25%, that's two fewer jobs a month, around $1,300.
- A bot missing two urgent calls a month (mistake 2) loses the jobs most likely to be accepted without a comparison quote. Two call-outs at $300 each is $600, plus the customers who'd have called you again next time.
- Nagging follow-ups (mistake 6) put off maybe one customer a month who'd otherwise have said yes: another $650.
That's roughly $2,500 a month of work lost to three fixable settings, before counting reviews or reputation. Your numbers will differ, and they're guesses unless you measure. The point is that these mistakes don't show up as a complaint; they show up as jobs that simply never arrive.
A 15-minute check of your own AI setup
Run these tests this week. Each takes a few minutes, and between them they cover all seven mistakes.
| Test | How | Pass looks like |
|---|---|---|
| Quote numbers | Compare your last three AI-assisted quotes against your price list | Every figure traces to your list or measurements |
| Emergency routing | Call your own number and say "water coming through the ceiling" | You get an urgent alert or the call transfers |
| Reply quality | Send your own enquiry form a specific question | The reply answers it with a real range or date |
| Photos | Scroll your website and profile gallery | You can name the job behind every photo |
| Reviews | Check who gets review requests | Every customer, automatically, after every job |
| Follow-ups | Look at the sequence settings | Three touches or fewer; stops on any reply |
| Website claims | Read the site as a sceptical customer | You could prove every claim on it |
Any row that fails is worth fixing before you add any new AI tool. None of the fixes needs new software; they're settings, prompts and ten minutes of honesty with yourself about what the AI has been saying on your behalf.
Further reads
- The First 30 Days of AI in a Trades Business — A steady first month of AI that avoids these traps.
- Free vs Paid AI Tools for a Small Trades Business — Which AI tools a small trades firm actually needs to pay for.
- Getting More Reviews After Every Job With AI Follow-Ups — Ask every customer for a review the honest way.
- How to Catch Made-Up Figures in AI-Drafted Proposals — Spot invented numbers before a quote goes out.
- What Tradespeople Actually Use AI For: 12 Real Jobs It Handles — Twelve jobs AI handles well for trades businesses.
- How to Set Up Missed Call Text Back for a Service Business — A simple fix for calls you miss while on a job.
- How a Plumber Can Save Five Hours a Week on Admin With AI — An illustrative time log for a sole plumber, six changes that add up to five hours a week, a four-week rollout and what it costs.
- How Electricians Use AI to Count Fittings From Drawings — How symbol search and trained detection count electrical fittings, which tools do it and what they cost, and the checks that stop a miscount reaching a tender.
- Nine Everyday AI Uses for Electricians Beyond Estimating — Nine electrician-specific jobs AI handles, from explaining inspection findings to landlords to logging intermittent faults, each with a worked example.
- Can AI Produce Accurate Estimates for a Small Building Firm? — Where AI speeds up a builder's estimate, where its prices drift, and a rate-library method that keeps the numbers yours.
- Can a Roofer Quote From Satellite Images? How AI Measurement Works — How aerial roof measurement reports are made, where they go wrong, and a two-stage way to quote from them without losing margin.
- Why Most Landscapers Haven't Adopted AI Yet, and Where to Start — The honest reasons landscaping firms hold off on AI, the three office jobs to start with, and a first-month plan for a four-crew firm.
- Can AI Replace the Home Survey for a Removal Company? — Which moves an AI video survey can quote without a visit, what it misses, and a hybrid survey that keeps most of the time saved.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Google Maps user-contributed content policy (fake engagement, selective solicitation of reviews, conflicts of interest); vendor documentation for AI receptionist and chat tools (checked September 2026).