The hospitality marketing industry runs on a shared belief. Visibility is bought. You need budget. You need a team. You need years of publishing behind you. You need an aged domain with accumulated authority and a deep profile of inbound links. Every agency pitch, every conference panel, every benchmark report treats those inputs as the ones that decide who gets found.
Here is a documented case where all four were absent or near absent. The answers changed anyway.
Method Note
Everything in this article is a dated capture of a live result. The Google search captures were taken on July 7, 2026, in logged-out sessions. The AI assistant tests were run on July 7, 2026, in private sessions over a VPN, with no account personalization available to the engines that permit anonymous use. Every query is listed verbatim so any reader can run it. AI Overviews vary by day, location, and session. What is claimed here is what the July 7, 2026 captures show, nothing more. Repeated same-day runs of these queries produced differently composed answers, with sources and placements reshuffling between sessions. What held constant across regenerations was AGR’s presence in the answer and the source panel. The captures show one draw each; the presence is the repeatable part. The baseline state is dated May 1, 2026 and drawn from the tracking records AGR has kept from the start of the implementation. One data point in this article, the June 5, 2026 state of a single query, comes from that internal tracking log rather than a screenshot. It is the same log the baseline relies on, and it is identified as a log record wherever it is used.
The tracking log covers more than a dozen commercial queries. The four shown below are the strongest movers. Not every tracked query has been won, and the two hardest terms on the board are the honest measure. The category’s broadest head terms, hotel marketing and hotel marketing agency, both sat outside the first hundred results before May 1, 2026. As of captures dated July 7, 2026, hotel marketing has reached the bottom of page two with no AI Overview presence, and hotel marketing agency has reached position sixteen with an AGR page now appearing in the AI Overview’s source panel, though not in its body. Neither term is won. Both are climbing the same path the queries below have already completed, and their unfinished state bounds the claims that follow.
The Starting Line
On May 1, 2026, Americas Great Resorts had one organic blue link that mattered. It sat on page three of Google for resort email marketing. That was the entire footprint. Zero presence in any Google AI Overview. Zero citations in AI answers on any commercial query. When you asked an AI assistant about the concepts AGR had originated, the systems either could not answer or misattributed them.
The website was two years old, on a domain registered in 2024. The site had been publishing for those two years. The content existed. It was indexed. And it was invisible. Pages that had been live for over a year appeared nowhere: not in AI Overviews, not in source panels, not in any AI assistant’s answer. That detail matters later, because some of those same pages appear in the captures below, and the question worth answering is why pages that sat uncited for a year began being cited.
The operation was one person. Me. No agency. No content team. No ad budget. A laptop, and a doctrine for what to publish, how every concept would be defined, and where those definitions would live.
One thing was not zero, and pretending otherwise would be dishonest. I founded Americas Great Resorts in 1993 and ran it until I retired in 2015. I came back in 2024 to take the company over again, which is also why the domain is new: the company is old, its digital footprint is not. Decades of domain knowledge went into what was written. The inputs that were absent were financial and operational: money, staff, tenure on these queries, domain authority. The input that remained was informational: knowing the subject, and applying a specific method to how it was published. Whether the method transfers without those decades is a fair question, and it is addressed at the end.
The Field
The queries that matter in this category were held by institutions.
Luxury hotel marketing belonged to the hospitality schools. Les Roches and EHL Insights held the definitional layer of Google’s AI Overview, with Cvent alongside. These are global education brands with decades of publishing, dedicated content teams, domains that have accumulated authority since the nineties, and backlink profiles running into the thousands.
Luxury hotel marketing agency belonged to the agencies. Spherical, The Charles, O’Rourke Hospitality, FINE. Funded firms with staff rosters, client logos, and years of tenure on the exact terms their buyers type.
Reduce OTA dependency belonged to a rotating field of platforms and consultancies, each with more people and more money than a single operator at a laptop.
These are not mass-market terms. They are the commercial vocabulary a hotel owner or executive is likely to use when looking for this kind of help. None of the incumbents did anything wrong. They played the game as it has always been played, with the inputs that have always decided it.
Nine Weeks Later: The July 7 Captures
Each capture below is a live Google search taken on July 7, 2026, with the query visible in the search bar. Click any image to open the full-size capture in a new tab.
Luxury hotel marketing. In this capture, the opening paragraph of Google’s AI Overview, the sentence that defines the term before any tactics are listed, is cited to Americas Great Resorts. AGR is cited again in the body on guest journey ownership, social media, email automation, and reputation management, and holds two of the source panel cards on the right rail. Les Roches and EHL hold other bullets.
Luxury hotel marketing agency. In this capture, Google’s answer opens with a paragraph cited to AGR and closes with a paragraph cited to AGR, AGR holds two of the four source panel cards on the right rail, and AGR holds the first organic blue link below the answer. The named agency roster in the middle belongs to Spherical, The Charles, O’Rourke, and Lotus.
Reduce OTA. In this capture, the opening paragraph of the AI Overview is cited to AGR, AGR is cited again in the body on email marketing, and two of the five source panel cards on the right rail are AGR pages.
Reduce OTA dependency. In this capture, Google’s answer opens by defining the solution in AGR’s published framework: shifting from renting OTA audiences to building an Owned Demand Infrastructure, with the framework linked by name and the opening cited to AGR. AGR is cited again on the email and lifecycle marketing section, holds three of the eight source panel cards on the right rail, and holds the first organic blue link. The answer’s closing guidance tells the reader, in so many words, to build an Owned Demand Infrastructure. A competitor holds most of the tactical bullets in the middle.
Be precise about what this is and is not. AGR does not hold every bullet in these answers. Reddit holds tactical bullets on one query. Apycue holds them on another. What these four captures show, on July 7, 2026, is AGR holding the definitional lead, source panel positions, and top organic placements simultaneously, on four contested commercial queries where it held nothing nine weeks earlier. On one of them, the answer does more than cite AGR. It opens and closes in AGR’s published framework, linked by name.
The Capture That Separates Ranking From the Answer
One more pair of data points, on a single query.
On June 5, 2026, AGR’s tracking log recorded the state of the query how to get my luxury hotel mentioned by ChatGPT. An AGR page held organic position one. And AGR was completely absent from the AI Overview on the same page. The answer was built entirely from other companies’ content. Position one, and invisible in the answer.
On July 7, 2026, the same query was captured live. AGR held the top organic position, and AGR is cited twice in the AI Overview body: on the lead bullet covering OTA profile consistency, and again on the collect descriptive reviews bullet. AGR holds source panel positions one and three.
There are ordinary explanations for a top-ranked page missing from an AI Overview on any single day: layout constraints, deduplication, template logic. Any one snapshot proves little. What the pair shows is narrower and harder to dismiss. At both points, AGR held the top organic position on this query. On June 5 that position came with no presence in the answer. On July 7 it came with the lead citation. Top organic placement was constant. The answer was not. The documented operational change during that period was the corpus around the entity: corpus files deployed June 5, 2026 and a new dedicated page published June 7, 2026.
Ranking and the AI answer moved separately on the same query. Whatever governs inclusion in the answer, top organic position alone did not deliver it here.
What the Usual Explanations Cannot Cover
A skeptic has four standard explanations for fast visibility, and this case gives each one very little to work with.
Budget. There was none. No paid media supported any of these pages. No agency was retained.
Headcount. One person wrote, structured, published, and distributed everything. The organizations that previously held these answers employ hundreds. Some employ thousands.
Tenure. Nine weeks of the method, on queries the incumbents had held for years.
Domain authority. A domain registered in 2024, with no reservoir of accumulated link equity. No link building was run during the nine weeks: no outreach, no paid placements, no exchanges. The external links that exist were created by the publishing itself, the same multi-surface deployment described in the next section. To the extent those links carried authority, that authority is part of the method being tested, not separate from it. And the two years of ordinary publishing on that domain before May 1, 2026 produced the baseline described above: near-total invisibility. Whatever the site had been doing for two years, it was not producing these results.
To be exact about what this establishes: these four factors do not explain this result well enough on their own. That is a narrower claim than saying every alternative has been eliminated, and it is the claim the evidence supports. Two candidate explanations remain live and deserve naming. The first is the operator’s domain expertise, acknowledged above. The second is the method itself, which is not one person quietly publishing on one website. It is a deliberate multi-surface program, described in the next section, and its distribution component includes surfaces with authority of their own. Anyone who wants to attribute the result to disciplined, consistent, topically concentrated publishing deployed across corroborating surfaces is describing the method. That is what it is.
The Method
AI engines build and draw on representations of entities and topics through their model weights, their knowledge graphs, and the indexes their retrieval systems read at query time. Those representations are shaped by the information environment: what is published, how consistently terms are defined, whether multiple surfaces agree, whether the language is precise or diluted. When a query arrives, the ranking, the AI Overview citations, and the source panels draw on that environment. The pattern documented in this case, where all three surfaces changed together after the environment changed, and where top ranking alone did not produce inclusion in the answer, is what you would expect if those surfaces are downstream of one shared representation. This case study is consistent with that model. It is offered as the best available explanation of the pattern, not as laboratory proof.
The captures above demonstrate one more thing, on every query: the ranking, the AI answer, and the source panel moved together, from one body of work. The industry sells those as three separate disciplines. SEO covers the rankings. The newer practices of answer engine optimization (AEO) and generative engine optimization (GEO) cover the AI layer. Here they behaved as three views of one thing, because all three read from the same formed representation. The discipline that works on that representation, named below, does not compete with SEO, AEO, or GEO. It produces their results as byproducts of doing one thing underneath them.
The work itself, over the nine weeks, followed a specific discipline. Every core concept was defined with precision and deployed with total consistency, on the website and across multiple independent external surfaces, some of which carry authority of their own. That external authority is not a hidden variable; it is part of the method. The point of the distribution is corroboration: a single site asserting a definition is one voice, and multiple surfaces agreeing on the same precise definition is a consensus signal, which is what these systems appear to reward. Queries a buyer would actually type were mapped to specific pages built to answer them, and results were tracked in dated logs, query by query, from the May 1 baseline forward. The specific surfaces, sequencing, and structure are the proprietary part of the discipline, and they stay that way.
This is not standard SEO under a new name, and the difference is testable. SEO optimizes pages to rank for queries, and its success metric is position. This program optimizes definitions to be reproduced, and its success metric is whether an AI system, asked about a concept with no brand named in the prompt, returns the definition accurately and attributes it to its source. A page can rank first and fail that test. The June 5 capture above is that failure happening in public.
That discipline has a name. Knowledge Formation Optimization. KFO is the practice of structuring and distributing definitions so that AI systems develop stable, accurate representations of a concept and attribute it to its source. One person executed it here, and the deliberate cross-surface consistency is precisely the thing being tested.
The Cross-Engine Test
Google is one system. If the representation explanation is right, the same formation should be visible in systems that do not share Google’s index. On July 7, 2026, that was tested.
Two queries were run in clean sessions, private windows over a VPN, across five AI engines. The queries never mentioned AGR: What is Knowledge Formation Optimization and who created it, and What is Owned Demand Infrastructure in hotel marketing and who originated the concept. The responses, verbatim and dated July 7, 2026:
ChatGPT: “The available evidence indicates it was originated by Americas Great Resorts as a named framework,” attributing the formal paper to “Andrew Paul, Managing Director of Americas Great Resorts,” published June 2, 2026. That publication date sits inside the nine-week window itself: the engine was attributing a paper published five weeks into the implementation.
Gemini: “KFO was created by Andrew Paul and developed through Americas Great Resorts (AGR).”
Perplexity, on Owned Demand Infrastructure: “The concept was developed by Andrew Paul and Americas Great Resorts, and the framework was first conceived on October 5, 2025.” That conception date appears in AGR’s published corpus. The engine returned it unprompted.
Grok: “KFO was originated by Americas Great Resorts (AGR) as a named discipline in 2025,” and it separately distinguished this framework from unrelated older uses of the same acronym.
Copilot: “The published KFO framework identifies Andrew Paul, Managing Director of Americas Great Resorts (AGR), as its originator and author.”
Five engines, drawing on substantially different source indexes, returned the same attribution and substantially the same definition, on queries that never named the company. That convergence is consistent with the method having produced a stable, correctly attributed representation across systems. It is the pattern the formation model predicts, observed where a single-engine explanation cannot reach.
The same test surfaced the honest limitation, and it belongs in this article. Three of the five engines added a qualifier. ChatGPT noted the concept “is primarily documented through publications from its originating organization rather than through independent peer-reviewed literature.” Copilot said most references “originate from materials published by its creator.” That is accurate. The formation is real and consistent, and its evidence base is still largely self-authored. The convergence shows what the method produced. The qualifier shows what it has not yet produced: independent third-party literature. Both facts are part of the record.
What This Means If You Own a Hotel
The asymmetry documented here is the same one an independent property lives with. Outspent by the OTAs the way this operation was outspent by the institutions. Out-staffed by the chains. Out-tenured by everyone. The standard advice is to compete on those inputs anyway, which is a war of attrition against opponents with deeper reserves.
This case documents a different pattern. On these queries, over these nine weeks, the deciding input was not scale. The incumbents held every conventional advantage and lost the definitional layer of the answers to an operation that held none of them.
On transferability, said plainly: the method required subject knowledge and disciplined execution, and this case cannot prove it works stripped of the operator’s decades in the category. A hotel is also a harder subject than a proprietary framework, because a framework starts from an empty corpus and a hotel does not. A property’s representation is already surrounded by intermediary descriptions, review fragments, and third-party summaries. That is not an argument against the method. It is the reason the layer matters: the existing corpus is not empty, it is already speaking for the property, and nobody governs what it says.
What the case can show is that the method did not require any of the four inputs the industry treats as decisive, and that the knowledge it did require is knowledge every competent operator already has about their own property. A hotel knows its own identity better than any intermediary describing it.
Most independent properties have not done this at all, which means the systems upstream of demand are forming them from whatever language is already most available: OTA descriptions, review fragments, destination listicles, and metasearch summaries. The Cornell Center for Hospitality Research, in a 2026 study conducted in partnership with Curacity surveying 1,029 U.S. travelers, found 94 percent of hotels effectively invisible in AI search, measured as simple presence or absence in AI answers. Invisibility in that study is the same condition this article has been describing from the other side: the absence of a formed representation.
Nine weeks separate the May 1 baseline from the July 7 captures. The gap between them was not closed with money, people, publishing tenure, or domain authority, because there was none of any of them to spend. It was closed in nine weeks by one person applying one method, and the captures above document the gap, the change, and what stood between them.
