Buying a Home Service Company? 21 Checks for Reviews and AI

Before you buy a home-service company, read the review record under its Google rating, which probably looks like its competitors'. At home-service locations connected to Cheers, the median location had 2.6% of a year's new Google reviews stop showing, and about one in seven lost 10% or more (a Cheers analysis of reviews posted July 1, 2025 to June 30, 2026). That record comes with the company.

Last verified: November 1, 2026.

In Cheers' weekly scans of Google's local map results across our home-service customers' markets (Sept. 28 to Oct. 31, 2026), 63.2% of rated competitor listings sat at 4.9 or 5.0. So the 21 checks look at what does differ: the review record, who controls each profile and domain, and whether AI names the company in its top towns. They run from before the LOI (letter of intent) through close and the first 90 days.

Deal teams and operating partners can run the pre-LOI checks this week without the seller, and an owner preparing to sell can run every check first. Deal counsel owns the reps and warranties and any access before close.

The 21 checks

Benchmarks come from dated sources in this article and aren't pass-or-fail rules; every "ask" is a question for the seller.

Checks 1 to 8 cover reviews (the record, other sites), check 9 covers markets, 10 to 13 profiles, 14 to 18 the domain and site and 19 to 21 AI answers. Each line gives when to run it, how, the benchmark and what's worth a question.

• 1, rating and review count (pre-LOI): the profile, plus Maps' top three for its main service. Most rated competitor listings are near-perfect; 12.5% sat under 4.5. Ask if it's under 4.5, or has far fewer reviews than those three.
• 2, review spread across locations (pre-LOI, diligence): public totals, with 12-month counts from the seller. At the typical company, the bottom half of locations got 11.8% of reviews (about 40% if even). Ask about any location with real revenue and almost no reviews.
• 3, reviews that stop showing (LOI and pre-close): two complete, comparable exports, checked for paging gaps and changed scope before treating a missing ID as a possible removal. The 90th-percentile location had 13.9% gone. Ask if one location lost 10% or more, starting with appeals, merges and moves.
• 4, star changes (diligence): the same exports, same ID with different stars. 5 to 1 was the most common change (63 of 283). Ask about several downgrades at one location, or a cluster in one month.
• 5, Yelp record (pre-LOI): each Yelp listing. ChatGPT's Yelp-linked cards averaged 4.34 stars where AI Mode showed 4.89 for the same 1,252 US businesses. Ask if one location's gap runs far wider than the rest, or one complaint keeps repeating.
• 6, BBB record (pre-LOI): rating, complaints and legal entity. Perplexity cited bbb.org in 81.8% of by-name answers, ChatGPT in 60.9%. Ask about open complaints, an entity that isn't the seller, or an old brand.
• 7, employer reviews (pre-LOI): Indeed, which Perplexity cited in 28.3% of by-name answers (its company reviews page in 24.2%). Ask about themes a customer would notice, like no-shows or pushy quotes.
• 8, how reviews are requested (diligence): interview, policy and request messages, measured against the FTC's review rule and Google's contribution policy. Ask about customer incentives, staff or relatives posting, staff quotas, requests to name staff, or asking only happy customers. Check employee attribution and recognition separately from those practices.
• 9, fake or duplicate listings (pre-LOI): Maps searches by trade, address and phone in each top town, with the pending Premium Home Service complaint as the reference. Ask about listings at the target's address, or numbers it doesn't recognize.
• 10, who owns each profile (diligence): owners, managers, business group and invitations. Only the primary owner can transfer it, so ask if that's a personal, ex-employee or agency account.
• 11, Google-held updates (diligence): the live profile against the seller's data. Google held its own values on 57 of 126 connected home-trade profiles (45.2%). Ask about a wrong phone, hours or category, or a closed label.
• 12, verification, restrictions and suspensions (diligence): status and open appeals, with appeals delayed again and no estimate. Ask about any restriction, suspension or open appeal.
• 13, messaging (signing to close): whether it's on, whose details are on file and who reads it; one reader reported an EIN request. Ask if nobody answers, or the details belong to the selling entity.
• 14, domain registrar and expiry (diligence): registrar, expiry, auto-renew and contact email, against ICANN's notice and redemption rules. Ask about expiry within a year, auto-renew off, or notices going to someone who left.
• 15, site admin and updates (diligence): CMS admins, hosting and who updates, with no benchmark. Ask if nobody can say who updates it.
• 16, machine-readable site (pre-LOI): the homepage with JavaScript off. 3.9% of US contractor sites refused our fetch (403), and 5.8% of readable homepages had under 150 words in raw HTML. A blank or blocked homepage is worth a question.
• 17, call-tracking numbers (diligence): every number, where it's published and who pays. 41.0% of checkable US ChatGPT-printed numbers weren't on the business's site when it ran call tracking, against 12.1% without. Ask about any number in AI answers nobody can place.
• 18, analytics, tag manager and Search Console owners (diligence): admin lists, with no benchmark. Ask if only the agency has access.
• 19, named in top revenue towns (pre-LOI, day 90): four buyer questions per town on ChatGPT, Gemini and Perplexity, and AI Mode separately. The target should be named about as often as the rivals beside it; ask if it's never named, or far behind the same rivals everywhere.
• 20, what AI says by name (pre-LOI): one by-name question per engine. Perplexity cited Angi in 50.3%. Ask about an answer built on an old brand or the previous owner.
• 21, the phone number AI prints (pre-LOI, day 90): each number against the site and tracking list. 17.6% of checkable US ChatGPT-printed numbers (66 of 374) weren't on the business's site, so a number nobody at the company recognizes is worth a question.

What to ask the seller for

More than half the checks need something only the seller has, so put these in the data request:

• A review export per location, from Google's Business Profile API or the seller's review tool, at LOI and again before close. Ask early, since a longer gap catches more.
• Each profile's primary owner, other owners and managers, and business group.
• Each domain's registrar account, expiry date, auto-renew status and contact email.
• Every call-tracking number and where it's published.
• Admin lists for the site, analytics, tag manager and Search Console.
• How staff ask for reviews, with copies of the messages.

Read the review record

Two exports taken weeks apart show what changed, the part a rating hides. Match them by review ID: a missing ID is a review that no longer shows, and the same ID with different stars is a star change. Google's review data documentation describes no history of earlier ratings, so one export can't show what moved. Spot-check two or three public profiles against them, and ask for edit history only if the seller's review software keeps one.

Across the 73 locations we measured, a quarter lost 0.6% or less and a quarter 5.5% or more; 20.5% lost none.

One-star reviews stopped showing about twice as often as five-stars: 4.1% of one-stars posted 2023 to 2025 at home-service locations, against 2.1% of five-stars.

Stars change too. From July 17 to Oct. 31, 2026, across every industry we serve and with one large brand excluded, the change we logged most often was five stars to one, a median 164 days after the review was posted (middle half 35 to 499 days). That's why the second export matters.

Cheers can't see why a review stopped showing, since Google's filters, a reviewer deleting it and a listing change look the same from outside. Treat a high rate as a question for the seller.

If reviews pile up at a few locations (check 2), our guide to auditing AI search visibility across locations checks each one on its own. If they've gone missing, start with the usual causes in why Google reviews go missing across locations.

Check the records AI reads besides Google

A homeowner who asks ChatGPT for a contractor can see a Yelp rating, and an engine asked about one company by name often builds its answer on the BBB, Angi or even Indeed, whose pages carry employer reviews from people who worked there.

Some gap between Google and Yelp is normal, since Yelp tells businesses not to ask for reviews and reviews it doesn't recommend don't count toward its star rating.

Find out who controls each Business Profile

Only the primary owner can transfer primary ownership, and a new owner or manager waits seven days before transferring ownership, removing users or deleting the profile (Google Business Profile Help). If the primary owner is a former employee's personal account, you'll need that person or a Google ownership request to get the profile back; our guide to transferring Business Profile ownership after an acquisition covers the handover step by step.

Compare each live profile with the seller's data field by field. Google says its updates come from "user reports and licensed content," and owners can accept or reject them (Manage Google Updates).

On Oct. 20, 2026, Search Engine Roundtable reported review appeals delayed again with no estimate, so a review that stops showing before close can stay gone well into your first months. The same day it reported Google testing verification with a government ID, so check verification status while the seller can still help.

On Oct. 28 the site posted a reader's screenshot of Google asking for an Employer Identification Number to turn on messaging, which Google hasn't announced; after an asset deal, expect to resubmit for the buying entity.

Find out who controls the domain and the site

ICANN's Expired Registration Recovery Policy requires renewal notices about a month and a week before a domain expires and one more within five days after, and registries hold most .com domains for a 30-day redemption window after deletion. Domains still lapse when the notices go to an inbox nobody reads anymore, so give the move into the buyer's registrar account a date on the closing checklist.

For the site, turn JavaScript off and reload the homepage, or open View Source, to see what a crawler or agent that doesn't run scripts gets.

Ask AI about the company in its top revenue towns

In each of the target's top 3 to 5 revenue towns, ask ChatGPT, Gemini and Perplexity the same four "who should I hire" questions for its main trade (12 answers per town), and check Google AI Mode separately. Start with "Who are the best [trade] companies in [town]?", add three a homeowner would ask, and keep all four word for word everywhere and at day 90. Then ask each engine what customers say about the company by name, and save every answer with its sources and phone number.

Compare the target with the two or three rivals that keep showing up beside it. Run the Cheers AI visibility grader for a repeatable API baseline, and keep app observations as a separate view of what buyers encounter. Use the same questions and method before and after close.

Read the phone numbers too. Counting repeats, ChatGPT printed about 6.4 times as many per answer after Oct. 9, 2026 in the Cheers Market Baseline (4.53 against 0.71), and some aren't on the business's own site (checks 17 and 21). Some may be tracking lines the business still pays for, so don't cancel one before you know where AI prints it, or you can cut off callers you never knew you had.

Questions about how the reviews were gathered

The FTC's Consumer Reviews and Testimonials Rule, in force since October 21, 2024, bars fake reviews, undisclosed reviews by insiders, incentives tied to positive reviews and suppressing reviews through threats or false accusations, with civil penalties for knowing violations. Google's contribution policy bars customer incentives, pressure "while on the premises" and asking only happy customers, and its review policy bars staff quotas and asking for content that names a staff member (Search Engine Roundtable, Apr. 17, 2026).

On May 11, 2026, the Justice Department, on referral from the FTC, and the Illinois Attorney General sued B.E.S.T. GDR, LLC, doing business as Premium Home Service, alleging more than 15,000 fake business profiles on Google Search and Maps propped up by fake reviews; the case is pending. On July 15 the FTC approved a final order against TruHeight, a supplement seller whose April complaint alleged several thousand five-star reviews written by employees. Neither is about a deal, but both describe practices that can come with the reviews a buyer takes over.

Our separate guide covers what the complaint alleges and how to report fake listings (check 9). In diligence, also ask which tools or vendors send review requests, and see our review compliance playbook for the asking rules in full.

Close and the first 90 days

The order matters more than the speed: ownership moves before anyone edits the public profile, and the AI answers you saved before close become the baseline for every fix.

Before close, inventory every profile and role, pick the buyer's owner accounts and business groups, and put the invitation date, seven-day wait, transfer window and domain moves in the closing checklist. Any earlier profile access goes through deal counsel, since owners can edit the profile and add or remove users on a business you don't own yet.

From close through day 7, the seller invites the buyer's accounts as owners and nobody edits the public profile. In days 8 to 14, transfer primary ownership where eligible and check from both accounts that it moved; days 15 to 30 are for removing old access, only after that check, and moving domains, analytics and tag manager access.

Days 31 to 90 can cover nonurgent cleanup. Fix wrong numbers, incorrect closed status and other customer-blocking errors immediately rather than waiting for this example schedule. Accept or reject Google-held updates, correct Yelp and BBB, and mark any branch you close or merge as closed on all three, with its number forwarding to a live location so AI doesn't send callers to a dead line. Rename the brand, if you're renaming it, after ownership moves.

Then rerun the same AI questions and phone checks against the pre-close save to see what the 90 days changed.

The point is one set of owners for every profile, domain and login, however many brands you buy. From a Cheers ad I recorded for operators (4:46 to 4:54 of a six-minute video).

Method

Customer data comes from Google Business Profiles connected to Cheers, which skew to established multi-location operators and may differ from a typical tuck-in; "companies" are root accounts and "locations" are individual profiles. Home-service locations and home-trade profiles have a home trade as their Google primary category: plumbing, HVAC, air conditioning, heating, electrical, roofing, water damage and building restoration, garage doors, handyman work, windows, fencing, waterproofing or landscape lighting. Every published customer figure has a base of at least 10 companies, with at least 10 locations behind both the base and the count, and no customer is named. All customer cuts ran Nov. 1, 2026.

Reviews no longer shown: 28,036 reviews Google returned with a review ID, posted July 1, 2025 to June 30, 2026, at 73 locations of 16 companies with 30 or more reviews that year; 11 of 73 (15.1%) lost 10% or more. The star comparison uses the same filter on reviews posted 2023 to 2025: 80 of 1,933 one-stars at 85 locations of 18 companies, against 64,950 five-stars at 109 locations of 19.

Star changes: 283 at 91 locations of 18 companies, July 17 to Oct. 31, 2026 (15 weeks, so earlier edits don't show), with one large brand excluded because its volume would dominate them. Five to one made up 63 (22.3%), at 36 locations of 10 companies, 40 of them at home-service locations. Review spread covers 11 companies with three or more home-service locations. Google-held updates are a point-in-time flag through Nov. 1, 2026 that doesn't say which field differs; the 57 flagged profiles sit at 11 companies.

Competitor listings: 15,795 from Cheers' weekly map scans in the markets of 10 Cheers home-service customers, with the customers' own and sister locations removed, so they aren't a national random sample. The website census read raw homepage HTML, with no JavaScript run, from 1,096 US home-service contractor sites drawn from Google's local map results, Cheers customers removed; 976 were readable as served.

Grader: the by-name citation shares (checks 6, 7 and 20) and naming rates come from the Cheers AI visibility grader, which asks ChatGPT, Gemini and Perplexity through each company's developer API, web search on, as fresh requests with no signed-in user, history or personalization. Each business gets four buyer questions built from its category and city and one that names it; until early October, that one asked what customers praise or complain about. Since September 23, ChatGPT and Perplexity also get the city as the asker's location, and in late September we reworded the buyer questions; from March 18 to June 8, ChatGPT answered the buyer questions for about a third of businesses in one combined request (since retired). The sample is 455 businesses, about 7 in 10 of them home-service, March 17 to October 30, 2026.

Market Baseline: every week it asks ChatGPT, Google AI Mode and Gemini the same 32 home-service questions in 100 metro areas, 60 in the US and 40 in Canada. The phone check sampled 900 numbers from ChatGPT answers collected October 2 to 31, 2026 and checked each against the named business's own website on Nov. 1; "not on the site" means not in the pages we checked, and nobody dialed them. The Yelp comparison covers Oct. 20 to 31, 2026. Grader and Baseline come from different panels, so we never chart them together.

Sources

• Work with review data, Google Business Profile APIs, updated August 28, 2026. No history of earlier ratings in the review data.
• Don't Ask for Reviews, Yelp Support Center, undated. Yelp's rule against asking for reviews.
• Don't Ask For Reviews: Why Yelp Does Not Recommend Solicited Reviews, Yelp, January 18, 2017. Not-recommended reviews don't count toward the star rating.
• Transfer primary ownership of a Business Profile, Google Business Profile Help, undated. Who can transfer, and the seven-day wait.
• Request ownership of a Business Profile, Google Business Profile Help, undated. Getting a profile back from an absent owner.
• Manage Google Updates, Google Business Profile APIs, updated August 28, 2026. Where Google-held updates come from and how owners answer them.
• Google Business Profiles Review Appeals Are Delayed Again, Search Engine Roundtable, October 20, 2026. Appeal delays with no estimate.
• Google Business Profiles Verification With Government ID, Search Engine Roundtable, October 20, 2026. The government ID verification test.
• Google Business Profiles Asking Businesses For EIN, Search Engine Roundtable, October 28, 2026. The reader's EIN request for messaging.
• Expired Registration Recovery Policy, ICANN, effective August 31, 2013. Renewal notice timing before and after expiry.
• The Consumer Reviews and Testimonials Rule: Questions and Answers, Federal Trade Commission, November 2024. What the FTC rule bars, and penalties.
• Maps user-contributed content policy, Google, undated. The bans on incentives, pressure and selective asks.
• Google Updates Reviews Policy To Restrict Staff Mentions & Solicitations, Search Engine Roundtable, April 17, 2026. The staff quota and staff-mention clauses.
• Department of Justice Files Complaint Against B.E.S.T. GDR, LLC, Doing Business As Premium Home Service, U.S. Department of Justice, May 11, 2026. The fake-profile complaint.
• FTC Approves Final Order Against TruHeight for Deceptive and Unsubstantiated Advertising of Supplements for Kids and Teens, Federal Trade Commission, July 15, 2026. The TruHeight final order.

Sources read November 1, 2026.

Dylan Allen-Arnegård is the CEO & Co-Founder of Cheers, the done-for-you local visibility partner for established home-service companies and PE-backed portfolios.