Contractor Review Generation: The Ask That Works Without Feeling Automated

Contractor review generation used to be a marketing chore. Since October 21, 2024 it has also been a compliance exercise, the day the FTC's Rule on the Use of Consumer Reviews and Testimonials (16 CFR Part 465) took effect, and in December 2025 the agency began sending warning letters citing civil penalties of up to $53,088 per violation. The two most common tactics in home services, a gift card for a five-star review and a survey that routes only happy customers to Google, are now the fastest ways to get a profile restricted.

The commercial pressure has not eased. BrightLocal's 2026 Local Consumer Review Survey, published February 11, 2026 from 1,002 US consumers, found 74% look for reviews written in the last three months and 47% will not use a business with fewer than 20 reviews. Recency is the scarce resource. A plumbing company with 300 reviews and nothing new since spring reads like a business that may have closed.

The problem is narrow: an ask that produces genuine reviews steadily, survives both rulebooks, and does not read like a machine.

Table of Contents

1. What Contractor Review Generation Has to Produce
2. The Two Rulebooks: The FTC Rule and Google's Review Policy
3. The Math: Ask Volume, Recency, and Why Averages Resist Repair
4. Five Ways the Ask Goes Wrong, and How to Spot Each One
5. The Ask: Timing, Script, and the Technician Handoff
6. A 90-Day Rollout, and What Not to Automate
7. Frequently Asked Questions
8. Ready to Build a Review Process You Can Defend?

What Contractor Review Generation Has to Produce

A review program has four outputs, not one: enough reviews to clear the credibility floor, a star average high enough to survive the filter consumers apply before clicking, a flow recent enough to look alive, and a documented process proving nothing was paid for. Most contractors optimize the first and ignore the rest. Review gating, any workflow that screens customers by sentiment and sends only the satisfied ones to a public review site, is how the fourth one gets lost.

BrightLocal's thresholds are specific enough to plan against: 68% of consumers require four or more stars and 31% will only use a business at 4.5 or higher; 47% rule out businesses with fewer than 20 reviews; 32% want reviews from the last two weeks. Rankings weight the same signals differently. Whitespark's 2026 Local Search Ranking Factors report, from 47 practitioners scoring 187 factors in November 2025, ranks a high Google star rating 6th for local pack visibility and recency 11th, while owner responses rank 122nd. Yet 89% of consumers expect responses. Replying barely moves rankings and strongly moves humans.

The Two Rulebooks: The FTC Rule and Google's Review Policy

Two sets of rules govern the ask and prohibit different things. The FTC rule is about money and authorship; Google's policy is about incentives and selection. A contractor can satisfy one and violate the other.

On the federal side, the rule's structure is what you audit against. Section 465.4 covers buying reviews, where purchasing one reaches anything of value, money, gift certificates, discounts or contest entries, exchanged for a review conditioned on its sentiment. Section 465.5 covers undisclosed insider reviews from officers, managers, employees or agents. Section 465.7 covers suppression, barring unfounded legal threats, intimidation or certain false accusations used to remove a negative review. The FTC's announcement of the final rule covers AI-generated reviews explicitly.

On the Google side, the incentive prohibition is broader. Google's prohibited and restricted content policy bars merchants from offering "incentives – such as payment, discounts, free goods and/or services" in exchange for reviews or removal of negative ones, and separately from "discourage or prohibit negative reviews, or selectively solicit positive reviews from customers." That second clause is the gating prohibition, and it has no FTC equivalent. Consequences run from content removal through account termination. Asking is explicitly fine: Google's guidance on reviews says businesses can ask customers to visit a Google link or scan a QR code.

Run this against your current workflow:

1. No incentive anywhere: not the invoice footer, the yard sign, the tech's script.
2. Every eligible customer gets the same ask, whatever happened on the job.
3. No branch of the automation routes a low satisfaction score away from the review link.
4. No employee, family or subcontractor review without a disclosed connection.
5. No legal threat, invoice hold or warranty leverage used to pull a review down.
6. Selected reviews on your site do not imply those are all of them.
7. Your request names your company and gives an opt-out.

None of this is legal advice; an hour with counsel is cheap against $53,088.

The Math: Ask Volume, Recency, and Why Averages Resist Repair

Recency is a throughput problem, and the required ask volume is usually five to ten times what a contractor sends today. Take a hypothetical six-truck residential mechanical contractor, with assumptions to replace with yours: 420 completed jobs a month, a mobile number on file for 78%, eligibility rules excluding 12%, and 10% conversion.

Cost assumes a $150 platform plus three coordinator hours at $28, identical either way. Fixed cost makes the sloppy version eight times more expensive per review. Compare the $8 to your own cost per inbound lead, the discipline behind calculating what a missed call costs. The 78% capture rate is the largest lever, fixable at booking.

Volume also governs how fast a rating moves. A contractor at 4.3 stars across 120 reviews holds 516 rating points. Reaching 4.5 means solving (516 + 5N) / (120 + N) = 4.5, so N = 48; reaching 4.6 needs N = 90. At 29 a month that is seven weeks and three months, and only if every new review is five stars.

It works the same way downward. One one-star review moves a 4.3 average across 120 reviews to 4.27; against a 25-review base it falls to 4.17, nearly five times the damage. Volume is insurance you can only buy in advance. And since 31% of consumers require 4.5, the gap between 4.3 and 4.5 is roughly a third of the market. Fix the complaints that recur in your review text first, usually an arrival-window problem rather than a technician one, which surfaces once the full first-call-to-invoice process is mapped.

Five Ways the Ask Goes Wrong, and How to Spot Each One

Review programs fail in five diagnosable ways, four of which look like success.

1. The gate. A satisfaction survey goes first, and only four- and five-star responses get the Google link. Detection: trace every branch of the automation; any path ending without the review link is a gate, whatever the vendor calls it.
2. The incentive nobody filed as one. A $25 account credit, a free filter, a quarterly drawing, a tech saying "mention me and I'll take something off next time." Detection: search every template for discount, credit, free, gift card and drawing, then ask three technicians separately what they say.
3. The batch dump. Someone exports two years of customers and sends 900 requests on a Tuesday. Reviews spike, then disappear as filters catch the pattern. Detection: plot reviews per week for the past year; a spike then a falling count is the signature.
4. The company tablet. Customers review on a device in your truck, so a cluster shares one account environment and reads as a conflict of interest. Detection: if the process involves handing anyone your phone, it is this.
5. The identical reply. Every response opens "Thank you for your feedback! We appreciate your business." BrightLocal found 50% of consumers are put off by templated responses and 81% expect a reply within a week. Detection: read your last 20 replies aloud; if the first six words repeat, that time is making a negative impression.

The Ask: Timing, Script, and the Technician Handoff

The ask that works is short, sent in the name of a person, and arrives while the customer still remembers the technician's name.

Capture the mobile number at booking, in the call script and the web form, not at invoicing where a third are gone.

Trigger on job completion in the field. The technician marks the job done, releasing the request after a set delay. That spreads requests across the week and avoids spikes.

Name a human on both ends. "Hi Karen, this is Dana at Harbor Mechanical. Mike finished the water heater swap this morning." Then the ask, then one link straight to the Google review form. No survey, no landing page: each extra screen costs completions.

Give them a subject without supplying the words. "It helps other homeowners to know what the job was and how it went" prompts a useful review. "Please mention our fast same-day service" is you writing their review, and it reads that way to everyone.

One follow-up three days later, then stop.

Ask everyone eligible, including the customer who was annoyed. Operators resist this, and it is what makes the program defensible. You will collect some three-star reviews, along with the specific language that makes a profile credible.

AI belongs on the back end: drafting reply variants from the job record, sorting review text into themes so you see whether complaints name the arrival window or the work, and flagging eligible jobs. It is not good at deciding what to say when a customer alleges damage. AI drafts, the person who ran the job edits, and no reply about safety or a legal claim goes out unread, the division of labor behind AI-drafted review responses.

A 90-Day Rollout, and What Not to Automate

Change one layer at a time and measure against a hold-out: a vendor's report on its own performance is not evidence.

Days 1 to 30, baseline. Count reviews received in the trailing 90 days and your mobile capture rate. Audit every template, script and printed piece for incentive language, then map the automation and delete any sentiment gate.

Days 31 to 60, one controlled change. Put the technician-triggered ask on half your crews and leave the rest unchanged. Measure reviews per 100 completed jobs for each group, so crew volume does not distort the result, and test send timing separately by alternating weeks.

Days 61 to 90, codify. Write down the eligibility rules, name an owner for the weekly reply rotation, and audit three numbers monthly: reviews in the trailing 90 days, reviews per 100 jobs, and average rating of new reviews.

Keep these out of the automation. Warranty callbacks and second visits, where the ask lands as tone deaf. Jobs with an unpaid balance or open dispute, where it looks like leverage. Any job a technician flagged. Property-damage jobs until the claim settles. Commercial and property management accounts, handled by a named person. And never let a tool compose the customer's review.

Process first, technology second. A review platform makes a good process faster, and a bad one faster too.

Frequently Asked Questions

Ready to Build a Review Process You Can Defend?

Review generation looks like a marketing task and behaves like an operations one: the eligibility rules, the job-completion handoff and the monthly audit matter more than which platform sends the text. ANtreliX is a strategy-first AI and business intelligence consultancy in North Myrtle Beach, South Carolina, working with home services companies locally and remotely to get the process right before adding automation. Talk to ANtreliX or see our case studies.