A viral marketing tip keeps surfacing: buy a domain called yourbusinessreviews.com, fill it with happy customer quotes, link to it from your main site, and watch ChatGPT recommend your business when a buyer asks about your category. The pitch is clean, the steps are cheap, and the framing is irresistible. It is also wrong about how AI search actually retrieves and weights sources. This is the technical breakdown of why the hack misreads ChatGPT's stack, what does drive citations, and the honest playbook for a reviews hub that earns real signal instead of pretending to be one. If you want the foundational frame first, start with How to Get Cited by ChatGPT in 2026 or run a free AEO audit to see where your site stands today.
The pitch usually runs like this. ChatGPT, when asked "is [your brand] any good," will quietly search Google for "[brand] reviews," find the dedicated reviews subdomain you set up, see only positive content, and pass that sentiment back to the user. Buy the domain for $12. Fill it with quotes. Free reputation engineering.
Every clause in that sentence is technically incorrect. The retrieval path is wrong, the bot model is wrong, and the assumption about how a 14-day-old domain ranks is wrong. The hack is selling a mechanism that does not exist.
ChatGPT Search uses a different retrieval architecture. OpenAI runs the user's query through one or more third-party search providers (Bing has been the dominant backend, with additional providers rotated in over time), rewrites the user's question into one or more targeted searches behind the scenes, and only triggers a web search when the question benefits from current information. A casual "is X any good" prompt may not trigger retrieval at all if the model has a confident answer from training data. When retrieval does fire, the rewritten query rarely looks like a literal Google search for "[brand name] reviews."
The mechanical fix is to stop optimizing for the wrong retrieval path. The retrieval that actually happens weights entity clarity, schema, and source trust over brand-name keyword matching on a thin standalone domain.
OpenAI runs three distinct crawlers, each with a different job:
A fresh reviews domain only gets crawled by OAI-SearchBot if it is indexed by the underlying provider, and only gets cited if it clears the same trust thresholds as every other source. A two-week-old domain with no inbound link graph and no editorial history clears nothing. The hack assumes crawling equals citing. The two are decoupled by a long list of ranking signals.
This is the cleanest failure mode. Search indexes treat domain age, citation graph, and topical authority as core ranking inputs. A standalone yourbusinessreviews.com lives in the same neighborhood as expired-domain spam, link farms, and PBN tactics that the major search engines have been suppressing for a decade. Your main brand domain almost certainly outranks a fresh reviews subdomain for any "[brand] reviews" query within a week of indexing. The hack has you competing with yourself and losing.
Google drew the explicit line here. Self-serving reviews of a business, published on that business's own site (or on a domain it controls), are not eligible for review rich results the same way third-party reviews are. Google's structured-data guidance is direct on this point and has been for years. The major LLMs inherit the same trust hierarchy: independent third-party review platforms (G2, Capterra, Trustpilot, Yelp, Clutch, DesignRush) carry weight because the editorial control sits outside the brand's hands.
A reviews subdomain you own carries the inverse signal. The honest test is whether a cold reader would assume the quotes were curated by an independent party. If the answer is no, the search engine and the LLM are running the same check and reaching the same answer.
When ChatGPT decides whether to name your brand in a category recommendation, the inputs are a stack of signals. The shape of the stack:
Notice what is not on the list: a separate reviews domain. The closest the stack gets to that signal is "third-party citations," and a domain you own is not third party.
A reviews or testimonials page on your main brand site is still useful. It is not a hack. The honest version, ordered by ROI:
The single biggest move is appearing on the platforms with editorial weight. For B2B software, G2 and Capterra. For local services, Yelp and Google Business Profile. For agencies and consultancies, Clutch and DesignRush. For ecommerce, Trustpilot and Yotpo with verified-buyer flows. Independent surfaces carry weight precisely because you do not control them. The constraint is the feature, not the bug.
On your main brand domain (not a separate one), build a real page that pulls together customer quotes, named clients, dates, case-study summaries, screenshots where permitted, and outbound links to the third-party platforms where those reviews originated. Mark it up with Organization or Service schema. Add Review schema only where the reviews are real, attributed, and accurate. Link to the page from your homepage and footer. Make it crawlable. Allow OAI-SearchBot, GPTBot, and ChatGPT-User. (Our own version of this lives at fifteenthmeridian.com/reviews.)
A mention in a category publication, a Substack newsletter, a podcast interview, or an industry directory is worth more to ChatGPT than fifty self-published quotes. The citation graph is the moat. Build the relationships, ship the guest posts, get quoted in roundups. Slow and unsexy. Also the part that compounds.
A curated llms.txt at your root, Organization schema sitewide, Article schema on editorial, FAQPage schema where you have visible Q&A, named author bylines linked to bio pages with Person schema. The full seven-step stack is in How to Get Cited by ChatGPT. None of these is novel. The novelty is in how few brands actually ship all of them.
Run scheduled queries through ChatGPT, Perplexity, Claude, and Google AI Overviews. Log citation share over time. We compared the six best AI-citation tools in Best AI Search Visibility Tools 2026. Manual spot checks (10 representative buyer queries a month, logged in a sheet) work as a free starting point. The Meridian15 AEO audit is the cheapest first step for confirming the foundation is in place.
To be fair, the hack contains one true observation buried under the wrong mechanism. ChatGPT does weight brand sentiment from sources it indexes. The volume of positive third-party signal does shift category recommendations. The error is in how to manufacture that signal. You cannot generate it by buying a domain. You generate it by being the brand customers actually want to write about on platforms you do not control.
The right framing is the inverse of the hack. Stop trying to game what ChatGPT sees. Become the brand that benefits when ChatGPT sees the truth.
The brands we audit who tried a variant of the hack share the same pattern. Time and attention got spent on a domain that never indexed, never ranked, never got cited. Meanwhile the main brand site was still missing Organization schema, still blocking OAI-SearchBot in robots.txt, still publishing marketing prose with no concrete claims for a model to lift. The opportunity cost is the real damage, not the hack itself.
The fix is the boring stack: third-party reviews on platforms with editorial weight, a Customer Stories hub on the root domain, schema, llms.txt, named authors, off-site citations. The brands shipping all of that compound. The brands chasing hacks reset every six months when the hack stops working.
The category of ChatGPT hacks is going to keep producing viral content for the next two years. Most of it will fail in the same shape: a real observation about how AI search behaves, paired with a mechanism that does not deliver. The brands that ignore the hacks and run the boring stack (third-party reviews, schema, llms.txt, named authors, real off-site citations) will be the ones cited when buyers ask AI for category recommendations. Whoever builds the citation graph first owns the recommendation share for the next two years.
If you only do three things this quarter: unblock all three OpenAI crawlers in robots.txt, get on the right third-party review platform for your category, and build one real Customer Stories hub on your root domain with Organization schema. Those three alone close the gap on most competitive sets we audit. The full retainer playbook lives at SEO and AEO services and the Calgary-specific applied version is at Calgary AEO services.
Probably not directly, if the reviews domain is clean and links back transparently. The bigger cost is opportunity. Time spent on a fresh domain that won't get cited is time not spent on third-party listings, schema, and content depth that will. Your main brand domain almost certainly outranks a 14-day-old subdomain for any branded reviews query inside a week of indexing.
OAI-SearchBot crawls public web content to surface it in ChatGPT Search citations. GPTBot crawls content that may be used in model training. ChatGPT-User fetches pages when a user inside ChatGPT clicks a link or asks for a page summary. All three are blockable in robots.txt. If any are blocked, the site is invisible to that retrieval path.
Sites that ship the full AEO stack (allowed crawlers, llms.txt, JSON-LD schema, FAQPage where Q&A is visible, named author, off-site entity coverage) typically begin appearing in ChatGPT browse-mode citations within 30 to 90 days. The on-site work caps out fast. The compounding lever is the off-site citation graph, which moves on a year-plus timeline.
For most categories, yes, with caveats. Trustpilot, G2, Capterra, Yelp, and Clutch carry weight precisely because the editorial control is outside the brand's hands. They also carry policy constraints (review-gating rules, response policies, dispute mechanisms) you have to operate inside. Pick the platforms that fit your category, run them honestly, and prefer verified-buyer flows over open-review systems.
Google does not award review rich results to a business that publishes reviews of itself on its own site. You can still mark up the page with Organization or Service schema, link out to the third-party platforms where the reviews originated, and treat it as a hub. The hub is useful for users and as a parseable signal for LLMs. It is not eligible for star-rating treatment in Google search results.
Audit which AI crawlers are allowed in your robots.txt and ship the fix if any are blocked. That is the floor. Anything else is wasted effort on a site the crawlers can't read. The free Meridian15 AEO audit reports exactly which AI crawlers are allowed or blocked, plus the rest of the citation-grade signal stack, in under 30 seconds.
No hacks. Just the stack.
Run a free AEO audit on your site, then talk to the team about an AEO retainer that ships the honest stack instead of the viral hack.
Talk to the AEO Team