HomeStars and the Trades AI Answer

Picture a homeowner in Burlington at nine on a Tuesday night, a wet patch spreading across the bedroom ceiling. She does not scroll ten blue links. She opens Perplexity and types: who are the best roofers in Burlington, Ontario. Six seconds later she has five names, each footnoted, and clicks the second footnote. It is a HomeStars listing. By the time she calls the next morning, the shortlist was built without her seeing a single roofer's homepage.

That sequence is now the front door to a high-value trades purchase, and it runs through a directory most contractors treat as an afterthought. Across nine Ontario cities, Formative Digital ran the same prompt against ChatGPT, Anthropic Claude, Google Gemini, and Perplexity, then logged every domain each cited. In roofing, one name showed up more consistently than any contractor or any other directory. What follows is why, and what a contractor does about it.

Why AI routes the "best roofer" question through HomeStars, not your website

AI engines route the "best roofer" question through HomeStars because a review directory answers a question your own website structurally cannot: is this contractor actually any good, according to people who paid them. An engine assembling a recommendation needs a trust signal neither it nor the contractor authored. Your homepage says you are excellent. Every roofer's homepage says that. The engine discounts it the moment it parses it.

Kevin Indig's Growth Memo analysis of 18,012 ChatGPT citations found that 44.2% of citations came from the first 30% of a page, meaning AI grabs the who, what, and where from the top and treats buried content as roughly 2.5 times less likely to be cited. A directory listing front-loads exactly that: business name, city, star rating, review count, and verification status, all in the first screen. Most contractor sites bury the verdict, if they state it at all.

What our Ontario citation scrape found about HomeStars

Our scrape found that HomeStars was the single most cross-engine-consistent directory in the trades. Formative Digital's May 2026 analysis of 1,732 AI-engine citations across nine Ontario cities asked each engine the same question: who are the top five best roofers in {city}, Ontario, with names, addresses, and sources. In roofing alone, homestars.com was cited 26 times, and it did something almost nothing else did: it appeared across two engines and eight of the nine cities, from Burlington and Hamilton to Toronto and Waterloo.

That eight-of-nine spread matters more than the raw count. ThreeBestRated was cited 37 times but only inside one engine, Claude, in six cities; Google's own properties took 35, again concentrated in ChatGPT; Yelp managed 13, Claude-only; the Better Business Bureau took 9 across two engines as a corroborating second source. HomeStars alone showed up in both Claude and Perplexity, nearly every city. For a contractor deciding where one hour of profile work pays off across the most AI surfaces, that coverage beats a tall bar in one column: ThreeBestRated wins the count, HomeStars wins the spread, and the spread is the higher-yield first move. This is first-party evidence, not a vendor claim, and the cross-engine split behind it is the subject of our study of the AI engine consensus gap.

How HomeStars verifies a review before it counts

HomeStars verifies a review by tying it to a real financial transaction and putting a human in the loop before publication. According to the HomeStars blog, the contractor must supply the paid invoice before a review publishes, a dedicated integrity team reads every incoming review, and each one must run at least 30 words. This is no fringe operation: founded in 2006 by Nancy Peterson and acquired by IAC in February 2017, HomeStars listed over two million companies by 2015, making it Canada's leading home-services review platform.

Every clause is a friction point, and the friction is the product. The invoice requirement stops a stranger or a competitor from manufacturing a review, the integrity team means a generated five-star paragraph gets read by someone whose job is to catch it, and the 30-word floor forces detail that is harder to fabricate at scale. None of this makes HomeStars perfect; it makes HomeStars expensive to game. That chain of paid job, produced invoice, human read, and length floor is what the Verified Badge encodes, and the resulting audit trail, which an engine never reads directly but recognizes in the platform's reputation, is exactly what an engine wants when it needs a verdict it can stand behind.

Which engines borrow that trust, and how often

Two engines borrow HomeStars' trust signal directly: Anthropic Claude and Perplexity. Every one of HomeStars' 26 roofing citations came from one of them. ChatGPT leaned on Google Maps and Knowledge Graph data instead, while Gemini wrapped its sources through vertexaisearch.cloud.google.com, its Vertex grounding layer, which appeared 83 times as a citation wrapper. The directory signal and the maps signal reach the homeowner through different doors.

This decides where you spend effort. If your buyers research on Perplexity or Claude, a verified HomeStars profile is close to mandatory; if they use ChatGPT, your Google Business Profile carries more of the load. Most contractors face buyers on all four, so the answer is rarely "pick one engine" and usually "cover the directory layer and the maps layer in parallel."

Why you cannot fake this signal on your own domain

You cannot fake this signal on your own domain because its entire value is that you did not author it. A roofer with a wall of testimonials reasonably asks why an engine would cite a third party instead of his own reviews page. The answer is the 2019 Google rule applied to a 2026 problem: self-published ratings are not independent evidence, and an engine assembling a recommendation needs that independence or it risks naming a business that simply markets well. The GEO study (Aggarwal et al., arXiv:2311.09735, KDD 2024) points the same way, finding generative engines surface a source up to 40% more often when the page adds citations, quotations, and statistics. The signal that lifts you is verifiable, attributable detail, not adjectives, and a HomeStars listing is a dense block of exactly that: third-party reviews, a Verified Badge, and a Star Score, all attributed to someone other than you. None of this guarantees a citation, and results depend on your industry, competition, and existing presence; what it does is move you from a source an engine must discount to one it can defend.

What a contractor does in the next 90 days

A contractor runs both plays at once, because the trap is treating them as a choice. The HomeStars profile earns you into the directory layer Claude and Perplexity cite; your own pages, written so the who, what, and where sit in the first 30%, give every engine a clean entity to attach that verdict to.

This maps to two of the twelve vectors Formative Digital works through with every client: Cite, earning verifiable third-party references, and Localize, making city-level signals unambiguous. No ranking dust, only verified trust and front-loaded facts. The same verification-and-spread logic governs higher-stakes verticals, which is why we ran it for how engines name dental practices across Ontario and which Ontario personal-injury firms AI engines surface, where the cost of an unverified recommendation is steeper.

Questions contractors ask us about HomeStars and AI

How does HomeStars verify its reviews? By tying each review to a paid job and a real person. The contractor must supply the paid invoice before a review publishes, a human integrity team reads every submission, and reviews must run at least 30 words. That combination makes the platform hard to fake at scale.

Is HomeStars legitimate and trustworthy for finding contractors? Yes, within its limits. It is Canada's leading home-services review platform, founded in 2006 and owned by IAC since 2017, and its invoice-plus-integrity-team model is more rigorous than an open star rating. The verification is genuine, which is why AI engines borrow its judgement.

How do you get a Verified Badge on HomeStars? You earn it: complete real jobs, then have paying customers leave reviews the platform validates against the invoice and the integrity team. The badge is the visible result of that process, not a setting you toggle on or a fee that skips the vetting.

Why does ChatGPT recommend directory sites instead of contractor websites? Because a directory answers the trust question and a contractor's own site cannot. Since 2019, self-published review stars carry no independent weight, so an engine reaches for a third party that already vetted the contractor. In our data, though, ChatGPT leaned more on Google Maps than on directories.

How do I get my business cited by Perplexity AI? Give Perplexity a defensible, retrievable reason to name you: a verified presence on a directory it already trusts, your own pages front-loaded with the facts in the first 30%, and consistent name, address, and phone across sources. Perplexity retrieves before it answers, so being easy to verify is the whole game.

Does having reviews on HomeStars help with Google and AI search? Yes, on both fronts. Verified third-party reviews are a recognized trust signal, and a strong HomeStars profile is one of the sources Claude and Perplexity cite directly for Ontario contractors. It is not a substitute for your own optimized pages; it is the half you cannot build on your own domain.