Ask every customer, not just the happy ones, within a few hours of finishing, with a short personal message that mentions the actual job and links straight to your Google review form. Send one reminder three to five days later. AI's job is writing the job-specific line from your notes, so each request reads as if you typed it.
Most tradespeople who don't get reviews aren't doing anything wrong in the request. They're asking too late, too rarely, or with a link that lands on a search page instead of the review box. The other trap is a popular shortcut that breaks Google's rules: sending "happy? leave a review; unhappy? tell us privately" messages, which is known as review gating. Google's policy says businesses must not "selectively solicit positive reviews from customers", so the request goes to everyone.
What Google allows when you ask for reviews
Google's content policy for Maps is short on this point, and worth reading once. In practice it means:
- No incentives. No discounts, free extras, prize draws or "$10 off your next service" for a review. Google calls this fake and misleading content and says it's strictly prohibited.
- No filtering. Ask everyone, or ask by a rule that doesn't depend on whether they seemed happy (every completed job, say). Don't discourage negative reviews.
- No steering the rating or content. The policy allows asking for genuine reviews "without offering incentives to do so or attempting to influence the rating", so "please leave us 5 stars" is out, and so is suggesting what they should write.
- No staff quotas. Google lists businesses "requesting that staff solicit a certain number of reviews" as prohibited. "Each engineer gets five reviews a week" is exactly that.
- No pressure on the spot. Don't hand the customer your phone and wait while they write one.
- No asking for particular content. Since April 2026 Google's rules also rule out asking for reviews that mention something specific, such as a staff member's name.
That last rule catches firms with more than one engineer, because naming the person who did the job feels friendly. Before, from an illustrative three-van electrical firm: "Hi [first name], thanks for choosing us! If you were happy with [engineer's first name]'s work, please mention him in a Google review so he gets the credit: [link]". Two problems: "if you were happy" filters by satisfaction, and "please mention him" asks for specific content. After: "Hi [first name], thanks for having us in to replace the consumer unit today. If you've got a minute, a Google review helps a small firm like ours: [link]". If the customer names the engineer anyway, that's their choice, and it's worth more for being unprompted.
If you'd like the reasoning behind the gating rule in more depth, what review gating is and how it gets businesses penalised covers it.
Getting the direct link to your review form
A link that opens the review box directly removes the step where most customers give up. In your Google Business Profile, choose Read reviews, then Get more reviews, then Copy to copy the link. The same screen offers a QR code, but only in a computer browser, not on a phone, so download it at your desk. Put the link in your request messages and the QR code on your invoice template and on a small card you leave with the customer.
Test the link the way a customer will meet it before any message goes out. Text it to your own phone, open it in a private browser window where you're not signed in to your business account, and check three things: it opens your business rather than a search results page, the review box appears after signing in with an ordinary Google account, and the business name and address are the ones customers know. Then check whatever your job software sends, because some tools use their own link rather than the one you copied.
When to ask, by type of trade job
Timing matters more than wording. Ask while the relief or satisfaction is fresh, but after anything that could sour it is sorted.
| Type of job | When to send the request | Reminder | Watch out for |
|---|---|---|---|
| Emergency call-out (leak, no power, lockout) | Same evening, once things are back to normal | 3 days later | Don't ask while they're still mopping up |
| One-day install or repair | Within 2 hours of finishing | 4 days later | A problem that shows the next morning; wait until the next day if that's common |
| Multi-day project (bathroom, rewire, extension) | When the snag list is cleared and they've signed off | 5 days later | Asking before snags are fixed invites a three-star "not finished" review |
| Routine service or maintenance visit | Same day, but only once a year per customer | None | Asking after every service annoys regulars |
| Commercial job | After sign-off, to your site or facilities contact | A week later | Company policies that stop staff reviewing suppliers; don't push |
The reminder should be shorter than the first message and give an easy way out. For example: "Hi [first name], just a nudge in case the link got buried. No worries if you haven't time. [link] [your name]". The "no worries" matters: it keeps the relationship intact for the customer who never gets round to it, and they're still a customer who might ring you next winter.
Two messages is the limit: the request and one reminder. A third reads as nagging, and a customer irritated enough to reply "stop texting me" is not going to leave you five stars.
Letting AI write the line that makes each request personal
Generic requests ("We'd love your feedback! Please review us") get ignored because they could be from anyone. One specific line about the job changes that, and it's a line AI can write from the notes you already keep. If your job software can pass the job notes into an AI step (or you paste them yourself), use an instruction like this:
Write a review request text for a customer of [firm], a one-person heating engineer.
Job notes: {{job_notes}}. Customer first name: {{first_name}}.
- Start with their first name, then one sentence about the specific job,
in plain words, using a detail from the notes.
- Then: "If you've got a minute, a Google review really helps a small business
like mine:" followed by {{review_link}}
- Under 280 characters. Friendly, not gushing. No exclamation marks.
- Don't mention stars, ratings, discounts or what to write.
- Sign off with {{your_name}}.Three outputs from different job notes (illustrative):
Notes: "Boiler no heat, replaced diverter valve, pressure reset, old lady,
very cold house, came 7pm"
Text: Hi [first name], glad we got the heating back on last night and the house
is warming up. If you've got a minute, a Google review really helps a small
business like mine: [link] [your name]
Notes: "Annual service, flue check, all fine, added inhibitor"
Text: Hi [first name], thanks for having me round for the boiler service today,
everything checked out fine. If you've got a minute, a Google review really
helps a small business like mine: [link] [your name]
Notes: "New radiator in loft conversion, 1200 double, moved pipe run
under floor, customer pleased with finish"
Text: Hi [first name], hope the new radiator is keeping the loft room warm. If you've
got a minute, a Google review really helps a small business like mine: [link] [your name]Check every output against the rules before you let it run automatically. Early versions of this kind of prompt often produce lines such as "We hope you're thrilled with your stunning new radiator! If you're happy, a 5-star review would mean the world!" That breaks two things: the tone doesn't sound like a heating engineer, and "a 5-star review" is an attempt to influence the rating. The "don't mention stars" line in the prompt is there because of exactly that output. Also check that personal details in your notes ("old lady", "very cold house") never appear in the message itself; the AI should use the detail, not quote your notes.
Automating the send from your job software, or without it
Pick the route that matches the tools you already use:
- Jobber: automatic Google review requests are part of its Marketing Suite ($99 a month add-on, included on the Plus plan), sent when a job closes, a visit is completed or the final invoice is paid. Choosing "final invoice paid" for multi-day projects avoids asking before snags are cleared.
- Housecall Pro: review requests go out automatically when you tap Finish on a job, and can be switched to after payment. By default it sends customers with Gmail addresses to Google and splits the rest between Facebook and your website; if Google is where you want reviews, change that split in the Reviews settings. Requests are personalised with the job details.
- ServiceTitan: its Review Response Generator drafts replies to reviews; check with your account manager how requests are set up on your package.
- No job software: a Zap or Make scenario triggered by "invoice paid" in your accounting software can pass the invoice description to an AI step and send the text. Or keep it manual: a saved text on your phone with a gap for one personal line, written in the van before you drive off. It takes 40 seconds per job.
For the Zapier route, do the task sum first. Zapier's free plan only runs two-step Zaps (a trigger and one action), and this needs at least two actions: the AI step and the text. On a paid plan, triggers and filter steps are free, the send uses one task, and an AI by Zapier step uses one, three or five tasks per run depending on the model tier you choose. At 35 requests a month with the three-task tier, that's 35 × 4 = 140 tasks, plus 35 for reminders: 175, comfortably inside Professional's 750. Check the tier on your own AI step, because it changes the sum more than volume does.
Whichever route you use, build in a way to skip a job. A realistic failure: an automation keyed to "invoice paid" sends its cheerful request to a customer who paid that morning but had rung the day before to complain that the new boiler was noisy. The first the firm hears of it is a one-star review quoting the text back: "They fixed nothing and then asked me for a review." The fix is a "no review request" tag or checkbox on the job, which you or the office sets for complaints, warranty call-backs and any job with an open snag, plus a filter step in the automation that stops tagged jobs. It costs nothing to run and only needs remembering once, at the moment the complaint comes in.
One heating engineer's month of review requests
Picture a hypothetical one-person heating engineer doing about 60 jobs a month: 35 boiler services, 20 repairs and 5 installs. Before any system, he asked occasionally, when a customer was chatty, and picked up two or three reviews a month.
His rule under the new setup: every repair and install gets a request; services get one only if the customer hasn't been asked in the past year. That's about 25 repair and install requests plus 10 service requests a month. The texts go out through his job software with the AI-written line, and a reminder follows four days later for anyone who hasn't clicked the link, if the software tracks that, or for everyone otherwise.
Three numbers go in a monthly note: requests sent, reviews received, and reviews by job type. Whatever the rate turns out to be, the breakdown is what teaches you something. If repairs produce reviews and services don't, stop asking after services and put the effort into repairs. If one kind of job keeps bringing in lower ratings, that's feedback on the work, not the request.
A filled-in note for an illustrative first month shows how that reads:
| Job type | Requests sent | Reviews received | Average rating |
|---|---|---|---|
| Repairs | 20 | 7 | 4.9 |
| Installs | 5 | 3 | 5.0 |
| Services (first ask this year) | 10 | 1 | 5.0 |
| Total | 35 | 11 |
Eleven reviews against the two or three he used to get, and most from repairs, where the customer felt the difference that evening. One review from ten service requests suggests those customers don't see a routine visit as worth writing about. A reasonable next step is to move the service ask to the visit where you find and fix something, and stop sending it after an uneventful check.
One awkward group needs its own rule: jobs for landlords and letting agents. The agent pays and books, the tenant lets you in, and the tenant's number reached you only so you could arrange access. Texting the tenant for a review uses their number for something they never agreed to. Ask the agent instead, once a quarter rather than after every job, as a commercial contact, and leave tenants out of the automation altogether.
Answering the reviews that come in
Reply to every review, briefly and specifically, including the good ones. Google's own tips favour short, personal replies over the same thank-you pasted everywhere. AI is useful for drafting replies, especially to negative reviews, where the first draft you'd write yourself is often too defensive. A three-star review might read: "Fixed the leak but left a mess in the airing cupboard." An AI draft you'd edit into shape: "Thanks for letting me know, and sorry about the airing cupboard. That's not how I like to leave a job. I'd like to come back and put it right; I'll give you a ring this week." Then actually ring. Good reviews deserve the same specificity. "Thanks for the review!" under every five-star post tells future customers nothing; "Thanks, [first name]. That diverter valve was a stubborn one, glad the house warmed up quickly" shows a real job and a real person. Allow for a delay, too: Google says owner replies can take up to 30 days to appear, so a reply that isn't showing the next day hasn't necessarily failed. Using AI to respond to Google reviews covers replies in general, and replying to negative reviews with AI goes into the difficult ones. Keep a person approving each reply before it posts; whether to let AI auto-post review replies explains why.
Signs your review requests are going wrong
- Reviews that all sound alike. If customers keep using your words ("prompt, tidy and professional"), your request is suggesting content. Take the suggestion out.
- Reviews that appear, then disappear. Google filters reviews it judges not genuine. Make sure customers post from their own devices, and never on your phone or tablet while you're there.
- Replies asking you to stop. You're sending too often or to the wrong people, such as regular service customers who get a request after every visit.
- Reviews landing on the wrong site. Check which link your software actually sends, especially if it routes customers between platforms by default.
- A dip in average rating after you start asking everyone. It can happen, and it isn't a bad sign on its own: you're now hearing from quiet customers as well as delighted and furious ones. A rating built from all your customers is worth more than a perfect score built from a few.
Once reviews are coming in steadily, they're also your best marketing copy; turning customer reviews into marketing copy shows how to reuse them with permission.
Further reads
- AI Review Management Tools for Small Businesses (2026) — Tools that manage requests and replies in one place.
- How to Handle a Fake or Unfair Review With AI's Help — What to do when a review isn't genuine.
- Chasing Unpaid Invoices With AI: Polite Reminders for Trades — The other follow-up every job needs.
- Seven AI Mistakes That Cost Tradespeople Jobs — Automation errors that cost trades work.
- How to Run a Customer Referral Programme With Automated Reminders — Turn happy reviewers into referrals.
- Jobber, Housecall Pro, and ServiceTitan AI Features Compared — How the big job platforms automate review requests.
- Free vs Paid AI Tools for a Small Trades Business — Which trade jobs free AI handles fine, the three moments it stops being enough, and the cheapest paid upgrades to try first.
- AI Tools for Roofing Contractors in 2026: Estimating and Leads — Roof measurement, instant estimates, lead replies and photo documentation tools for roofers, with current prices and three stacks by firm size.
- AI Service Plan Reminders for Heating and Cooling Customers — Clean the due-date list, time reminders to the heating and cooling seasons, write messages about each customer's actual system and let AI sort replies.
- Following Up Silent Quotes With AI: A Decorator's Playbook — Four follow-up plays for decorating quotes that go quiet, from the day-two check-in to the day-thirty close-out, with AI drafting each one from your notes.
- Photo Checklists and Quality Control for Cleaning Teams Using AI — Set checkpoints a camera can prove, a fixed photo set per clean, an AI pass that flags which jobs need a supervisor, and clear privacy rules for clients' homes.
- Following Up Unapproved Repair Estimates With AI — A step-by-step way for repair trades to chase open estimates with AI-drafted messages that explain the fault honestly and allow partial approval.
- What Tradespeople Actually Use AI For: 12 Real Jobs It Handles — Twelve everyday trade jobs AI handles well, from voice-note job sheets to invoice chasers, each with a real-looking example and the mistake to watch for.
- 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 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.
- Can AI Give Customers a Ballpark Price for Decorating Work? — How a decorator can let AI give honest ballpark prices from a room-by-room grid, and which jobs it should refuse to price.
- 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 on fake engagement; Google Business Profile help on review links and tips for getting reviews; Jobber review request feature pages; Housecall Pro Reviews help article; ServiceTitan Titan Intelligence page.