Google’s AI Mode showed text ads on 29.45% of commercial queries in a new SE Ranking study, less than a year after ads began appearing in AI-generated search answers.
Ads appeared on 14,733 queries in the study, which analyzed U.S. commercial searches where text ads could appear, excluding product carousels.
AI Mode ad adoption accelerated.SE Ranking said ads began appearing in AI Mode responses in late 2025. By mid-2026, nearly one in three commercial queries in its data set showed a text ad. SE Ranking said the real ad rate may be higher because AI Mode results are inconsistent across sessions.
Two advertisers often appeared together. Most AI Mode ad blocks included more than one advertiser. SE Ranking found that 71.1% of ad-triggering queries showed two ads in the same AI Mode response, while 28.9% showed only one.
CPC best predicted ad visibility. Higher-cost keywords were much more likely to trigger AI Mode ads.
Ad presence was 24.33% for keywords with CPCs below $2.
It rose to 32.45% for keywords between $2 and $10, then jumped to 53.56% for keywords at $10 or more.
SE Ranking said search volume and keyword difficulty did not show the same relationship with ad frequency.
Ad presence varied across categories. Pets had the highest rate, with ads appearing on 72.38% of analyzed keywords. Healthcare had the lowest, at 2.64%.
Higher-ad categories were typically lead-generation markets with clear paths to paid conversions.
Lower-ad categories were more likely to involve informational or YMYL intent, where commercial demand may be lower or Google may be more cautious.
Advertisers were rarely cited. Buying an AI Mode ad didn’t make advertisers more likely to be cited as a source in the same response. Only 11.53% of advertiser domains appeared among the cited sources for the keywords they advertised on. At the URL level, overlap fell to 1.95%.
This held true even after comparing advertisers with similar non-advertising domains based on domain strength, backlinks, referring domains, and organic visibility, SE Ranking said.
Organic overlap was limited. Advertisers also rarely ranked organically for the same keywords. Only 2.32% of advertised URLs ranked organically for the queries where their ads appeared. At the domain level, overlap increased to 15.35%.
About 85% of advertisers didn’t appear in organic results for the same keywords where they showed AI Mode ads, according to the study.
Why we care. Buying visibility in AI Mode doesn’t increase your chances of being cited or ranking organically. Treat AI Mode ads, cited sources, and organic rankings as separate visibility channels.
About the data. SE Ranking analyzed 50,032 commercial keywords across 20 niches, averaging about 2,500 keywords per niche. The data reflects U.S. AI Mode results collected on June 30. SE Ranking noted that ad behavior may change as Google expands AI-specific formats.
Google’s AI Mode increased citations to google.com by 8.4x in about two months, making it the No. 2 cited domain in Profound’s tracking.
The increase came almost entirely from Google Business Profiles and Product Knowledge Panels, according to Profound. Those Google-hosted cards now appear within AI Mode answers for many local and product searches.
Google cards moved up. AI Mode now surfaces Google Business Profiles as inline panels for local-intent queries.
These panels can display a business’s hours, photos, location, and reviews before users reach the company’s website. Profound said the shift makes the Google-hosted profile the first page many users see.
Local categories led. The change was strongest in industries where local intent drives revenue, including:
Hospitality and travel
Home services
Restaurants and dining
Real estate
Healthcare
Product panels gained ground. Product searches also shifted toward Google-hosted results. Queries about comparisons, compatibility, or specifications increasingly surfaced Product Knowledge Panels instead of direct links to ecommerce or brand websites, Profound said.
Why we care. Your Google-hosted profile may shape a user’s first impression before they ever reach your website. Missing hours, outdated photos, incomplete information, or poor reviews can cost you the click before users visit your site.
About the data. Profound tracked AI Mode citation share from April 15 through June 30, analyzing more than 32 million google.com/searchviewer instances.
Salesforce found 20% of sales coming from agents marks a signpost.
60% of companies use agents live in production, and 3 out of 4 companies invest in AI agents.
To figure out how ready B2B sites are for agentic visitors, I teamed up with David Kaufman, founder of Siteline, and I analyzed how agents scan websites and where they get stuck. The answer: most sites are agent-ready, but there is one critical breaking point.
Agents don’t read websites like humans. They receive a task, search the web, fetch pages, extract facts, and cite the sources they used. A page can persuade a human and still fail an agent if the facts are hard to find (opacity), hard to fetch (machine-readability), or hard to cite (access friction).
AI agents turn websites from showrooms into barcodes.
How we looked at agent behavior:
The agent had to find the official site itself. We did not provide starting links, eg to homepages.
We gave agents three buyer-related tasks for 100 B2B products: find pricing/features, integrations, and security/compliance. We ran each task five times to measure the impact of the probabilistic nature of LLMs.
We weren’t comparing whether or not the information existed somewhere on the web; instead, we measured whether the agent could reliably answer from the vendor’s own site.
The moment a prospect looks at pricing, they stop browsing and start comparing. High buyer intent, bottom of the funnel. That makes pricing the hardest and most important test of whether a vendor site can serve agents directly.
Pricing also sits in a triangle of 3 “wants” that good pricing pages need to satisfy:
Companies want to control pricing disclosure.
Buyers want fast comparison.
Agents need clear, fetchable, citable facts.
When AI agents try to retrieve pricing, they get stuck much more than for security or integrations.
Pricing/features produced 77% of all third-party citations.
If you wonder whether that’s because some B2B companies don’t publicly show pricing, you’re only half right.
2. Hidden pricing is only part of it
Hiding prices forces agents to look elsewhere, but published prices do not fully solve the problem. Among pricing prompt runs where the vendor did not disclose a real price, 45% cited at least one third-party source. The other 55% stayed on first-party citations, usually by saying the vendor required contact sales or did not publish a concrete price.
Even when the vendor showed a numeric public price, agents still cited at least one third-party source in 18% of runs, suggesting price can be on the page but still be hard for the agent to extract, trust, or cite cleanly.
You can try to hide your pricing, but you better make sure no one else knows and writes about it. Once it’s “out there”, it’s too late. If you have complex pricing methodology, the best way is to explain it clearly and make it accessible to agents.
Some pricing pages are visible to humans but not reliable enough for agents to parse and cite. You can’t always trust your eyes.
3. Agents fail for three reasons
Agents fail to retrieve pricing from a brand for 3 reasons: opacity, machine-readability, and access friction.
Example of an agent struggling to retrieve Zendesk’s pricing and pivoting to third-party sources.
Pricing opacity simply means the brand doesn’t publicly disclose the price, or it’s vaguely packaged. Opacity explains elevated fallback, meaning agents have to rely on third parties for information.
Machine-readability describes the situation when prices exist, but agents still do not confidently extract them. Machine-readability explains fallback despite disclosed pricing. Machine-readability fails when the price is hard to extract because of page structure, JavaScript, calculators, toggles, screenshots, PDFs, or ambiguous tables.
Access friction is what most people expect to be the problem with agents. The agent hits fetch failures, rate limits, blocking, or unreachable pages, making agent runs more costly.
Access errors were not the main reason agents left first-party sources, but when they happened, they were severe. They appeared in only 7% of all runs. In pricing runs, access errors pushed third-party fallback to 77%, compared to 17% without access errors.
The impact of errors on agent run cost (tokens, web searches, fetches, retrieves, time) is significant when comparing the 90th with the 10th percentile in our study:
Cost: 4.4x
Token: 4.7x
Time: 2.0x
Brands don’t pay that bill directly, but it is a useful proxy for friction. The harder your site is to retrieve, the more work an agent has to do before it can answer from your page. If your pricing page is blocked, slow, hard to fetch, or hard to parse, the agent has two choices: spend more work on your site… or get the answer somewhere else.
4. The fallback web is messy
Fallback occurs when agents have to rely on third-party sources rather than first-party sources as a result of the three failure modes. This is the biggest risk because third-party information is spotty and beyond your control.
Agents do not fall back to one clean source category. They reconstruct pricing from a mixed web of explainers, directories, app stores, partner pages, and low-trust aggregators.
Key stats from the 580 pricing third-party citations:
52% were editorial (blogs, media articles, comparison guides, explainers, and other article-style pages).
46% fell into the directory category (review, procurement, and software-listing sites such as G2, Capterra, Vendr, Tekpon, and similar domains).
2% from broader ecosystem pages (app stores, marketplaces, partner pages, and integration directories tied to another platform).
The examples show the risk of missing pricing transparency and agent stumbling blocks on your site.
Example journey:
Here, failure mode indicates the reason the agent failed to obtain on-site / first-party pricing information.
5. How to make your site agent-proof
An agent-proof pricing page is how you keep the agent quoting you instead of a directory like Vendr. The fixes map to the three failure modes.
Disclose the fact (opacity)
Publish real prices in text for every self-serve tier. If a tier is genuinely custom, say what drives the number instead of “contact sales.”
Keep plan names, prices, limits, and features on one canonical pricing URL, and point every other mention back to it.
Mark legacy plans clearly so third-party content can’t keep stale tiers alive.
Make the fact extractable (machine-readability)
Put prices in crawlable HTML. Many agent fetches never run JavaScript, so a price rendered client-side is invisible. In testing, prices in server HTML got read in under a second; a JavaScript-only price got missed.
Add schema.org Product and Offer markup with price and priceCurrency. This single lever moved a page from 73 to 93 in the readiness test.
Explain usage-based pricing in text, not a calculator-only widget.
Let the agent in (access friction)
Allow AI crawlers in robots.txt (GPTBot, ClaudeBot, PerplexityBot, Google-Extended). Check you aren’t allowing Googlebot while blocking them.
Don’t block server-side AI fetches on pricing pages. Access errors hit only 7% of runs, but they push fallback from 17% to 77% when they do.
Keep the price early in the DOM and the page light. A 1 MB pricing page taxes every agent and pushes cheap runs to route around you.
Fix opacity and machine-readability first; they drive most of the fallback. Then run the query yourself, “Find all pricing and features for [product],” and measure it with the skill below.
This post first appeared on the author’s website and is republished here with permission.
SEO has long focused on what a page says. Increasingly, it also needs to account for how that information is presented.
As Google gets better at understanding page layout, structure, and functionality, visual semantics is becoming an important part of how search engines interpret webpages.
What is visual semantics?
Visual semantics is a meaning model for segmenting, classifying, and understanding documents by working alongside textual semantics.
Google is changing how it interprets web documents, shifting from “web text” to “web layout” to better identify real expertise, uniqueness, and originality by giving more weight to the functional components of a webpage.
Google’s Quality Rater Guidelines cite “human effort and involvement” as one of the most important quality principles, with “design effort” identified as one aspect of that evaluation.
Webpage layout has always been an important part of SEO, dating back to Google’s Page Layout algorithms. Those early algorithms focused primarily on ad placement and simple document-ranking signals, unlike today’s more sophisticated approaches to understanding webpages.
Why Google is paying more attention to page layout
Google has introduced newer inventions and patents that highlight the importance of understanding webpage layout. Most webpages are no longer built with only prose or simple text-over-text layouts. Instead, they contain much denser information.
Every 10 to 20 pixels can introduce a new interaction point, engagement element, clickable module, comparison unit, or dynamic component designed to help users.
That’s why some of Google’s leading engineers, including those who have worked on Gemini and AI Mode, are also associated with newer inventions such as Structured Information Cards and layout-aware multimodal document understanding.
As a result, it needs systems that can understand how different card types are structured, including product cards, hotel cards, real estate cards, trip cards, credit card cards, and other information cards.
In other words, modern search engines must understand not only the text on a page but also the layout, hierarchy, visual relationships, annotations, and functional meaning of each structured information block.
A citation from Google’s “Layout-aware Multimodal Document Understanding” patent
Why layout matters for search engines
Understanding structured information cards and layout-aware document interpretation requires neural networks, and possibly a new type of LLM, that can “verbalize” web documents with annotations and high-confidence citations.
Google can’t reliably rank a flight booking website, a credit card application aggregator, or similar platforms without understanding the data embedded in these documents.
Much of that data is presented through uniquely designed card structures, comparison modules, tables, and interactive layouts rather than plain text.
Below is an early example of document layout understanding from Microsoft called ViPS, which Google has also cited.
Both approaches are closely related and rely heavily on HTML to determine which text belongs to each section, component, entity, or visual block on a page.
With the rise of embedding-based algorithms, concepts such as “chunking” have become widely discussed in the SEO industry.
However, many discussions about text or document chunking miss a critical point: Chunking isn’t only a linguistic process. It’s also a layout-aware and structure-aware process.
If a document isn’t visually segmented and structurally understandable to search engines, the content itself becomes harder to interpret. In that case, it doesn’t matter how many entities, predicates, triples, or entity relationships you include, or how accurate they are.
Search engines still need to understand where each piece of information belongs, how it relates to the surrounding elements, and which visual or functional component gives it meaning.
In modern search, information quality alone isn’t enough. Information also needs to be presented within a layout that helps machines understand its boundaries, hierarchy, context, and purpose.
Google explained this concept through “centerpiece annotation,” describing visual annotations that help its systems better understand a document.
Martin Splitt from Google said the “centerpiece annotation” represents the “primary content” of a webpage.
Later, documents disclosed during Google’s antitrust case showed that centerpiece annotation was also used to classify and rank news documents.
The centerpiece annotation was primarily limited to about 400 characters, though those documents also reveal several other noteworthy details.
For example, below you can see how Google extracts the centerpiece annotation from HTML. The sentence is interrupted by unnecessary HTML elements, such as Facebook, email, Twitter (X), and Google+ share buttons.
In the next example from Google’s DOJ documents, proper HTML structure prevents share-button boilerplate from interrupting the centerpiece annotation, allowing Google to extract the content correctly.
What visual semantics looks like in practice
Below is a simple SEO case study. Although it involved 19 changes, the biggest ranking improvement came from one simple adjustment: moving a calculator component from the bottom of the page to the top, making it the centerpiece annotation.
The results of that change are shown below.
Metric
Previous
Current
Increase / Change
Success %
Total clicks
3.47 million
4.53 million
+1.06 million clicks
+30.5%
Total impressions
84.1 million
167 million
+82.9M impressions
+98.6%
Average CTR
4.1%
2.7%
-1.4 percentage points
-34.1%
Average position
8.9
8.5
Improved by 0.4 positions
+4.5% improvement
This project closely connects visual semantics and textual semantics because it’s a programmatic SEO case study involving more than 100,000 pages.
At that scale, even a small sentence edit, component update, or layout adjustment is multiplied across every URL. That’s why Google re-crawled the entire website after the layout changes and why impressions and clicks increased afterward.
The project is a converter website that ranks for queries such as “2m to cm” and millions of similar numeric and metric variations. In this type of search environment, more than 10,000 competing websites provide essentially the same data and the same answer.
These websites have the same topical coverage and factual accuracy. The competitive advantage doesn’t come from providing a better answer because “1 meter to cm” has the same value everywhere.
It comes from retrieval cost, document understanding efficiency, internal PageRank distribution, and how clearly the answer is presented for Google’s initial ranking systems.
In these types of queries, you can’t differentiate yourself by changing the answer. You differentiate yourself by changing how the answer is structured, annotated, prioritized, and visually presented.
That’s why changing the centerpiece annotation caused Google to reprocess the layout, rerank the pages, and further improve the site’s rankings.
What is the cost of retrieval, and how does it relate to visual semantics?
“The cost of ranking a document” can’t be higher than the “cost of not ranking a document.” I introduced this concept years ago in one of my conference presentations. Google cares about search quality, but its systems also weigh quality against cost. If a website costs more to process than its quality justifies, Google will look for an alternative.
Google reduced the HTML file size limit to 2 MB and carried out large-scale deindexing following the December 2025 core update.
At the same time, it sent a clear signal to websites that scale AI-generated content without meaningful human effort. Google appears less tolerant of practices it accepted for years, and its indexing decisions are likely to become even more selective.
Retrieval costs increase when a webpage doesn’t clearly explain itself or fails to demonstrate sufficient relevance and responsiveness, especially around the “centerpiece annotation.” Google’s Content Warehouse API leak suggests the company truncates documents and predicts quality based on initial signals. If a document doesn’t meet relevance and responsiveness thresholds during those early evaluations, it won’t be considered a candidate.
During Google’s antitrust trial, Pandu Nayak, then Google’s vice president of Search, explained that Google doesn’t run its most computationally expensive algorithms on every webpage because it lacks sufficient click data. Instead, it first evaluates core topicality signals to determine whether a page is worth indexing and keeping as a candidate.
Nayak also explained that RankBrain-like algorithms are expensive to run, so Google reserves them for results that have at least one click, demonstrate strong topicality, and include annotations that justify the investment in crawling, rendering, evaluation, and further processing.
In other words, classifying documents by their layout, components, and structured information cards is a more efficient way to reduce retrieval costs while improving search quality.
Today, most large-scale content publishers rely on AI to generate more text. Far fewer invest in front-end and back-end systems that improve user engagement, interaction, and document understanding.
That distinction increasingly separates low-quality and high-quality sources. Low-quality sources primarily scale text. High-quality sources scale systems, layouts, components, structured information cards, and user interactions that help both users and search engines understand content more efficiently.
Below is Google’s concept of website representation vectors.
Google classifies websites using visual and layout-related embeddings and features to determine whether they resemble expert, apprentice, or amateur sources.
“For instance, the website classifications may include a first category of websites authored by experts in the knowledge domain (for example, doctors), a second category authored by apprentices (for example, medical students), and a third category authored by laypersons…”
How does Google’s helpful content system relate to visual semantics?
The helpful content system is a classifier that identifies which websites genuinely provide helpful information or meaningful engagement and which only imitate usefulness without fulfilling the searcher’s underlying intent.
Much of the SEO industry’s analysis of the helpful content system has focused on textual features. Early discussions centered on keyword stuffing, gibberish content, or adding “unique information” to improve information gain. However, many of the system’s algorithms appear to focus on the function and type of a source.
Google first classifies websites by their type rather than their content quality. That means the same content can rank differently on an affiliate website than it does on an ecommerce website.
So how does Google distinguish among affiliate sites, aggregators, service providers, ecommerce sites, and SaaS platforms? The answer is visual semantics. What a page can do, or can’t do, is largely determined by its layout and page components.
The biggest distinction between relevance and responsiveness comes from engagement, not understanding.
Google created systems such as neural matching to align the entity type and entity ID in a query with the most relevant documents. In simple terms, if the entity in the query doesn’t match the entity in the document, that page becomes less likely to rank. This is primarily about relevance.
Relevance alone isn’t enough. A document may rank because it’s relevant, but if it doesn’t support meaningful user actions, such as purchasing, comparing, ordering, reviewing, filtering, or watching, it isn’t responsive to the user’s actual task.
That’s why the helpful content system shouldn’t be viewed only as a system that evaluates page text. It also evaluates page function. A helpful page isn’t simply one that contains relevant words. It’s one that helps users complete the action, decision, or information-seeking task behind the query.
Google reinforced this idea by adding “misleading functionality” to its spam policies after the Helpful Content updates. A page can appear helpful by imitating a function without actually providing it.
For example, a page may suggest users can compare, filter, calculate, book, review, or purchase something even though those functions don’t genuinely exist. In those cases, the page may appear functional to both users and algorithms, but it isn’t truly responsive to the user’s task.
Google doesn’t classify websites only by page layout and design. It also appears to apply result-type constraints within the SERP. For example, a query such as “best women’s glasses” may return listicles, ecommerce category pages, product grids, videos, and commercial guides in the same results page.
To satisfy multiple search intents, Google can apply diversity constraints that limit how many ecommerce pages, listicles, videos, or other result types appear together.
Google’s DOJ documents include functions such as “max_total” and “BlogCategorizer,” which show how Twiddlers can classify results and limit the number of pages from the same cluster, category, or source type.
A similar annotation appears in the Google Content Warehouse API leak through the “WebrefFatcatCategory” module, which assigns categorical weight to a result.
In other words, Google doesn’t simply rank documents individually. It also classifies, clusters, and constrains results based on page type, source category, and categorical diversity. As a result, a page may be relevant enough to rank but still be limited by the overall composition of the SERP.
Even when a generated ranked entity list, such as a “best products” page, ranks successfully, it doesn’t rank simply because it’s a blog article. It ranks because it functions as a commercial resource. It helps users compare, evaluate, filter, review, and move closer to a decision. In that sense, Google can rank nonfunctional content when it effectively serves a functional category.
Viewed through this lens, “helpful” in the context of the helpful content system is closely aligned with “functional.”
The following case study demonstrates this principle. We moved identical content from an affiliate website to an ecommerce website, supported it with an integrated topical map, and saw rankings improve almost immediately.
The content itself didn’t change. What changed was the function, context, and source type surrounding it. By placing the same information within a more functional, commercial, and task-oriented environment, Google interpreted the document as more useful for the user’s search activity.
How is click data used to rerank search results through visual semantics?
Google increasingly understands the purpose of a webpage through its layout, not just its text. As a result, click data is aggregated according to the type of source. Many SEOs assume that long clicks, or longer dwell times, signal quality.
However, that’s not always true, according to Google’s research. Depending on the category, shorter dwell times can indicate a successful experience, while longer sessions may signal an “engagement trap.”
Below is Google’s reranking model, which applies different ranking and rank-modification models based on user behavior captured by its tracking components.
Google also uses the concept of the “Life of a Click” to help engineers understand how search ranking algorithms interpret user behavior.
Taken together, these systems suggest that click data becomes a more meaningful classification signal when interpreted alongside a webpage’s design rather than through text alone.
Classifying documents by their visual structure can be more efficient than analyzing millions of documents, billions of word tokens, co-occurrences, named entity resolutions, attribute extractions, and value corrections.
If certain document layouts consistently generate stronger user satisfaction, Google can classify those pages as more helpful or functional. It can then use those signals to identify other documents with similar layout patterns, component structures, and interaction models.
This means topical authority doesn’t come only from a topical map that defines which topics to cover. It also comes from understanding which page layouts, component structures, information cards, comparison modules, and functional designs best match each topic, query, and search activity.
A proper topical map shouldn’t define only entities, attributes, predicates, and contextual relationships. It should also define the page type and functional layout needed to satisfy both relevance and responsiveness.
This leads to the concepts of coverage and domain-level classification. The following three examples illustrate this approach.
The first example is AudioToText.com, a sub-brand built around a single topic.
GSC Metrics of Audiototext.com. The third-party Semrush data is shown below.
Despite covering only one topic across 12 languages, or 13 pages in total, the site continues to grow in search visibility for three reasons:
Its exact-match domain reinforces relevance.
Its visual semantics improve responsiveness.
It earns its first clicks quickly, allowing Google to run more computationally expensive ranking systems sooner.
Click satisfaction from the other language versions may also reinforce the English version through cross-lingual information retrieval.
Google can use webpage layout understanding and chain-of-reasoning to classify AudioToText.com as a “no-signup transcription tool” and rank it in AI Overviews. This suggests Google isn’t only reading the text. It’s also interpreting the page’s function, visual annotations, and interaction model.
In other words, Google can use agentic retrieval based on visual signals to understand what a page does and determine whether it deserves to rank for a specific query.
The Audiototext.com’s single-page topical map representation with the fundamentals are below.
The webpage was designed with minimal text while placing its primary conversion element, the content upload component, above the fold.
If that component were moved lower on the page or made smaller, rankings would likely decline, and text changes alone wouldn’t be enough to recover them.
Another example is attorneys.lexinter.net, which ranks primarily through a subdomain because its core content was moved there together with a filtering engagement component.
The primary domain didn’t meet the required thresholds, but moving the content to a subdomain with additional functional elements produced better results.
The same subdomain testing approach also worked for Pricelisto.com. Although most of the design and content remained the same, we added functions and annotations related to purchasing, comparing, examining, and reviewing.
Those functional additions made the pages behave less like passive content and more like task-completing commercial resources. As a result, the site avoided filters associated with the Helpful Content System.
The improvement didn’t come from changing the text. It came from changing how the document functioned, how users interacted with it, and how clearly Google understood the purpose of each page component.
Search engines try to reduce retrieval costs by avoiding computationally expensive algorithms whenever possible. As a result, domains affected by historical or domain-level signals may not receive a completely fresh evaluation immediately.
Testing on a subdomain can give Google a clearer reason to reprocess documents, reevaluate their layouts, and run more advanced evaluation systems. That makes it easier to determine whether improvements come from new designs, functionality, annotations, or document structures rather than from the historical state of the primary domain.
How is visual semantics related to the future of search?
Google is experimenting with fundamental changes to search results, including replacing the traditional search bar with new interfaces.
One example is its Jan. 29 patent, “AI-generated content page tailored to a specific user.” The patent describes generating a landing page that uses visual segmentation, annotations, and generative AI to satisfy a user’s query.
The patent places significant emphasis on “landing page score,” using click data and explicit user feedback signals
In other words, Google can use visual semantics not only to rank web documents but also to construct new types of search results.
Google’s patent work is often complemented by its research. For example, the paper “Neural Design Network: Graphic Layout Generation with Constraints” explores how systems can understand, classify, and even generate webpage layouts to improve search performance.
This suggests that layout isn’t only a design consideration. It can also serve as a retrieval, classification, and ranking signal.
Google’s multimodal document understanding also connects to its latest announcement, Google Embedding 2, which uses generative neural networks to understand and vectorize text, images, videos, audio, and documents.
This matters because different versions of the same web document can be compared through their vector representations. Doing so makes it possible to evaluate how well Google understands layout differences, visual structure, and document-level meaning.
In other words, layout changes aren’t merely visual. They can also produce different vector representations, which may affect how a document is understood, classified, and retrieved.
Below is Google’s example of the neural network process for understanding page layouts. The centerpiece annotation that helps classify a webpage as an ecommerce category page, product page, or SaaS page comes from these types of labeling systems.
In the future, Google could apply these same principles to construct its own landing pages from multiple search results.
The patent shown below also illustrates how Google could adjust SERP features based on an entity’s primary attributes. That suggests search results aren’t simply ranked and displayed. They can also be reorganized, redesigned, and presented as dynamic interfaces based on the entity, query intent, and available document structures.
Google’s “Search result ranking and presentation” patent explains which types of visual representations best match different attribute-value pairs when generating SERP features and ranking webpages with similar characteristics
Centerpiece annotation and query processing
Google classifies and augments queries differently from how people naturally think about them. That means one of the most important parts of creating a topical map is understanding search terms the way Google’s systems do and augmenting them accordingly. This process is called query semantics. Below is an example of query augmentation from ChatGPT.
In this example, we searched for “best search engine optimization information sources,” and GPT expanded the query as follows:
Best SEO information sources: search engine optimization resources Google research, patents, SEO blogs
If you perform a search in ChatGPT, open the Network tab in Chrome DevTools, filter for XHR requests, and inspect the JSON file associated with the https://chatgpt.com/backend-api/conversation/6a* path. Look for search_model_queries, which shows what the system actually searches for.
Google also has a patent called query augmentation, shown below.
The patent is attributed to engineers, including Krishna Bharat and Anand Shukla. These names are significant because they also appear on patents and systems related to AI Overviews and AI Mode.
For example, the “Search with Stateful Chat” patent includes query augmentation as one of its steps, and its terminology and inventors overlap with this system.
The centerpiece annotation is the primary visual annotation that reflects a webpage’s purpose, function, and context. The context created through the augmented query needs to align with that centerpiece annotation.
The following case study shows how I classified query variations and their contexts across different document types, each with a distinct purpose, function, and visual structure, for a local service directory.
Let’s use “air conditioner” queries as an example. Each query variation should be matched with the appropriate page type, layout, and function.
Experience queries require a forum-style layout. For a query such as “How do I repair my AC?” the intent is experience-based. A forum structure works best because users expect real problems, answers, troubleshooting paths, and personal experiences. This content can also live on a subdomain to separate experiential content from the main commercial website.
Local service queries require a directory page. For “Air conditioner installation in [City],” the intent is local and service-oriented. The best page type is a local directory or listing page with providers, service areas, ratings, contact options, and conversion elements.
Price queries require a hybrid layout. For “air conditioner installation prices,” the intent is both informational and commercial. The page should provide an immediate answer with average prices, cost factors, and price ranges while also presenting local providers, comparisons, and quote-related elements.
Instructional queries require an informational layout. For “How to install an air conditioner,” the intent is instructional. The page should minimize local service elements and instead focus on a step-by-step guide, required tools, safety considerations, visuals, and practical instructions.
In short, a topical map should define not only which topics to cover but also the appropriate layout, components, and page function for each search activity. The following example shows some of the early results from this project after classifying query augmentation models for different query variations.
Early GSC results for the same brand.
If there’s no need for a separate page for the [Local], [Service], [Forum], or [Instructional List] intent, we simply prune it. If other pages are too similar, we merge them.
As a result, the number of pages decreases along with retrieval costs, while PageRank concentration and relevance per document increase. Below are four closely connected components:
Mock-up design in draw.io.
Production design in Figma.
Topical map for different query types.
Content brief aligned with the Figma and draw.io designs.
Early on, we defined the topical authority formula as:
Historical data x Topical coverage
Later, we expanded it to:
Historical data x Topical coverage ÷ Cost of retrieval
Today, I’d extend the formula with one additional factor:
((Historical data x Topical coverage) ÷ Cost of retrieval) x Right visual annotations
Even if you have the lowest retrieval cost, the highest topical relevance, the broadest topical coverage, strong accuracy, the longest duration of satisfied click data, and positive historical performance, none of it matters if the centerpiece annotation is wrong or the page isn’t functional.
Google’s ranking system largely functions as a decision tree. If the first decision-making layer rejects a website, the later evaluations, tests, and reranking processes won’t occur.
To maximize your chances of ranking from the start, visual annotations should be optimized just as carefully as the page’s text, images, and links.
Below is a conceptual model of this system.
A website consists of “letters, pixels, and bytes.” Data2Website is the process of turning a dataset that Google’s algorithms favor into a website by combining textual and visual semantics through those letters, pixels, and bytes.
The example above shows how a local law firm benefited from a topical map, semantically optimized content briefs, specific sentence structures, and visual design decisions.
The Semrush results below show the impact on the firm’s local rankings.
We previously applied the same principles to another ecommerce website.
If you examine the screenshots closely, you’ll see that the same principles carry over from an ecommerce design to a local service provider.
For every attribute within an entity-seeking query, such as “best law firm in Houston” or “birth test kit prices,” you can classify those attributes within the query network and organize them according to their importance.
Some attributes require review components, while others require directly commercial components.
Below are two design examples from the sibling websites Morethanpanel.com and StreamingMafia.com. Their above-the-fold and below-the-fold sections are structured similarly, covering different types of user engagement and functionality.
The above-the-fold area is often referred to as the macro-context because it contains the main content. Google’s Quality Rater Guidelines use the concept of main content to emphasize the importance of relevance, accuracy, and completeness in this section.
The below-the-fold area corresponds to what Google’s Quality Rater Guidelines describe as supplementary content, which we refer to as the micro-context. This section typically contains less important attributes and most internal links.
The next example shows the mock-up design and the distribution of factual content, opinionated content, structured content, and unstructured content.
Google doesn’t always prioritize factual or opinionated content, or structured versus unstructured content. Instead, it evaluates these characteristics based on how the search query is augmented. To improve language relevance, we distribute different types and formats of content using different visualization, verbalization, commercialization, and contextualization techniques.
The following example applies the same approach to the second website in the same industry, together with its topical map, content briefs, and authorship rules.
Algorithmic authorship can be explained through the research paper “Are LLMs Reliable Rankers?” It means writing content according to predefined sentence structures and rules. For example, the research shows that the “Rank anything first” framework increased rankings by 20% to 60%.
The system evaluates which words should follow one another to determine how relevance changes. It performs retrieval within a generative retrieval system and identifies the entity-attribute-value triples that best improve relevance. In the example above, “material” is selected as the attribute and “steel” as the value because they strengthen relevance within that context.
Structured content: Supports attributes such as symptoms, advantages, and benefits.
Unstructured content: Supports concepts such as definitions, processes, and importance.
Visualization: Presents content using the appropriate semantic attributes.
Commercialization: Adds functional components that help users complete their tasks.
Contextualization: Maintains relevance by aligning content with the query.
Verbalization: Converts visually important information into text that LLMs and search engine crawlers can understand.
Depending on the query, Google may prefer opinionated and unstructured content, factual and structured content, or other combinations supported by different visualization, commercialization, contextualization, and verbalization techniques.
The following example from the online dating industry shows how different webpage components can improve relevance and responsiveness at the same time.
The next examples illustrate different ways to visualize content.
Comparing these two sections, you’ll see that one answer is highly factual, while the other, distinguished by a different background color, is more conversational and opinion-based.
We can create a Q&A component and add opinion-based content as forum-style discussions at the bottom of the page.
We can also ask users questions and let them contribute answers through voting, allowing those responses to be verbalized into content that is continuously updated.
Below is what we call the preceding question component. It reframes the original question using a semantically similar concept and gradually shifts the content from factual to more opinion-based.
The next example shows a horizontal tab component that distributes internal links to related headings, increasing contextual coverage.
The following Semrush data shows the early and later results for the URLs we modified.
The patents and research behind visual semantics
At this point, we’ve introduced the key concepts, definitions, and website examples needed to explain visual semantics.
We could explore these examples, processes, and implementation details in much greater depth, but every conceptual discussion begins with understanding where Google is heading.
Many of Google’s advances in query semantics, visual semantics, Gemini, and AI Search are driven by two influential engineers: Dr. Marc Najork and Michael Bendersky. They are among Google’s most frequently cited researchers in recent years and have played major roles in shaping the company’s AI-related direction.
They are also listed as inventors on the Layout-Aware Document Understanding and Structured Information Cards patents.
Another important contributor is Alexander Grushetsky, who identifies himself as the founder of RankLab, Google’s internal end-to-end ranking platform.
He’s worth mentioning because he’s frequently cited alongside Bendersky and Najork in foundational patents and research papers.
Grushetsky also worked with Bendersky and other Google engineers on item-ranking models based on item types, attribute sets, and attribute values. We’ll explore what RankLab represents in more detail another time.
Today’s search engines and large language models increasingly rely on visual semantics as part of their vectorization and embedding-based ranking systems.
Even the original Transformer research described extending these ideas to web documents and their layouts.
WebRef vectorizes webpages using not only their text but also their visual layout, page components, HTML structure, and overall document context.
Whether your rankings depend primarily on external PageRank, branded search demand, or internal signals such as semantics, a page’s visual context still carries ranking weight alongside its textual relevance.
If you apply standard SEO playbooks to a travel website, you’ll likely exhaust your budget with little to show for it.
Most SEO advice is written for sectors operating in a search ecosystem where organic text listings still dominate most user journeys.
In travel, the rules of search are entirely different because Google isn’t just a search engine. It’s a direct transactional competitor, a visual aggregator, and a gatekeeper to visibility.
Winning in this space requires accepting that what works elsewhere will fail here. To build a search strategy that actually drives bookings, travel brands must abandon standard organic best practices and instead master unique challenges:
Intermittent, non-linear user journeys and highly fragmented user intent.
Travel search is driven by data, not webpages
In almost every other industry, the ultimate goal of SEO is to secure a high-ranking organic text listing.
In the travel sector, particularly for high-intent queries involving hotels, flights, and activities, the traditional “blue link” is functionally dead (and becoming even less relevant as AI, AI Overviews, and other LLMs compete for user attention).
If a user searches for “flights from London to Rome” or “boutique hotels in Edinburgh,” the top of the search page is entirely occupied by Google’s own interactive search tools.
Below these tools sit local map packs, sponsored advertisements, and, in the UK and Europe, the massive “Find results on” directory boxes mandated by antitrust regulations.
Instead of just writing content, successful travel SEOs focus on feed management and entity optimization.
For accommodations and hotels, this means treating Google Hotel Center with the same priority an ecommerce specialist gives their primary website. You must ensure that your real-time pricing feeds, inventory levels, and tax calculations are integrated directly with Google’s API.
Travel entity optimization requires meticulous calibration of your Google Business Profile. Google’s local algorithms categorize hotels and attractions based on physical attributes rather than editorial text.
Whether your property appears in a filtered search for “dog-friendly hotels with free parking” depends entirely on your structured attributes, customer review sentiment analysis, and precise location coordinates.
In travel, the search engine is a database, and your primary job is to format your data so the database can display it without friction.
Traditional search campaigns operate under the assumption that organic traffic must land directly on a domain you own to have any measurable value.
In the travel sector, this insular approach ignores how modern searchers, especially younger audiences seeking visual reassurance before booking, actually behave.
Google has adapted to this shift by turning its results pages into visual aggregators that pull content directly from external social networks.
For queries driven by discovery or curation, such as “best rooftop bars in Soho” or “hidden beaches in Cornwall,” Google often serves dedicated “Short Videos” carousels and “Perspectives” feeds directly within the main search results.
These features extract short-form video content from TikTok, Instagram Reels, and YouTube Shorts, allowing users to watch authentic, crowdsourced footage without ever visiting a traditional website. With search engines actively indexing public business posts from platforms like Meta, social media updates are now ingested into search layouts and AI Overviews.
Travel SEO is no longer confined to the boundaries of your content management system. To capture this visual search real estate, digital teams must treat social media profiles as distributed landing pages.
The standard content marketing advice for B2B or consumer services is to build comprehensive, long-form informational guides. The theory is that if you write a 4,000-word article covering every possible detail of a destination, you’ll build topical authority and guide the reader smoothly down a conversion funnel.
In travel SEO, this approach often yields high bounce rates and very few conversions. Travelers don’t plan trips in a linear fashion, nor do they want to read extensive blocks of text on mobile devices while walking down a busy street.
Travel planning is highly visual, emotional, and fragmented across different devices and moments.
A user searching for “things to do in Cornwall” is usually looking for quick inspiration, geography-based grouping, and immediate utility. If they land on a page with lengthy introductory paragraphs and dense blocks of prose, they’ll return to the search results to find a simpler resource.
Travel content must be built for rapid utility rather than high word counts. Instead of writing long essays, your content layouts should rely on interactive, modular components, such as:
Tabbed interfaces that allow users to toggle between “Itinerary,” “Cost,” and “Best Time to Visit.”
Embedded, interactive maps that show the physical proximity of recommended locations.
Bite-sized, structured lists that clearly state opening hours, entry prices, and booking links.
Structuring content this way also makes it much easier for Google to extract your data for its AI Overviews and visual carousels.
Instead of trying to keep users on a long page of text, the goal should be to provide structured, high-value answers that Google can easily parse.
When your site behaves like a functional tool rather than an online magazine, users are much more likely to bookmark your pages and trust your brand when they’re finally ready to book.
Winning at travel SEO requires a fundamental shift in perspective. Success can’t be measured solely by standard keyword rankings or overall organic traffic levels.
It must be evaluated by how effectively your site appears across Google’s SERP features, how reliably your data feeds communicate real-time pricing, and how quickly your landing pages answer highly specific, visual questions.
By stepping away from generic SEO advice and focusing on technical data feeds, regulatory price compliance, and modular user experiences, travel brands can secure a distinct competitive advantage in one of the most crowded landscapes on the web.
Hidden gem publishers have 1.7x higher audience affinity than major media outlets, even though they attract 130x less traffic. If that makes you rethink your 2026 earned media strategy, it should.
Fractl’s latest SparkToro-backed research suggests most digital PR teams still build media lists using outdated SEO metrics such as domain authority and traffic. (Disclosure: I’m the co-founder of Fractl.) In AI-driven search, entity authority shapes brand visibility. A narrow media-list strategy is increasingly misaligned.
If you’re still learning about generative engine optimization (GEO), entity authority is the cumulative signal created when your brand is repeatedly associated with credible sources across influential channels in your niche.
In plain terms, your brand visibility now relies on which sites mention you, which audiences engage with those sites, and whether your brand keeps showing up across trusted nodes.
How authority has evolved in today’s AI-centric landscape
For years, digital PR teams built media lists the same way SEO teams built link targets: prioritizing domain authority, traffic, and referring domains. That made sense when backlinks were the primary goal.
But AI-driven discovery is changing what “authority” actually means.
When ChatGPT, Gemini, Perplexity, or Google’s AI experiences synthesize information about your brand, they’re not just evaluating whether your website is optimized. They’re pulling from the broader web of sources that repeatedly mention and contextualize your brand.
A placement on a smaller, highly relevant industry publication can sometimes do more to reinforce your brand’s entity authority than a broad mention on a much larger site.
That shift reflects what we’ve seen across years of earned media campaigns spanning major publishers like The Wall Street Journal, The New York Times, and CNBC, alongside niche publications such as PCMag, Men’s Health, and Travel + Leisure.
Today, those campaigns often extend far beyond backlinks, generating TV coverage, Reddit discussions, podcasts, YouTube commentary, and other third-party mentions that reinforce brand authority across search, social, and AI-driven discovery.
Which non-mainstream media publishers drive the strongest influence?
We wanted to understand which publishers and platforms actually drive influence with the audiences brands want to reach. That question led to this study.
We used SparkToro’s audience affinity data as a relevance layer on top of traditional earned media and SEO metrics. Rather than starting with the largest publications in each category, we began with the audiences brands want to influence: decision-makers, buyers, and practitioners across eight industries.
We then analyzed where those audiences spend time across websites, YouTube channels, podcasts, social accounts, and community-led platforms.
After separating mainstream publishers from vertical-specific outlets, we compared audience affinity against traditional signals such as domain rating, organic traffic, and referring domains.
The study revealed a major blind spot: “Hidden gem” publishers had 1.7x higher audience affinity than mainstream media outlets, despite attracting 130x less traffic.
Many highly relevant publications would never surface on a traffic- or authority-sorted media list, even though they may play an important role in shaping the topical associations AI systems use to understand and describe brands.
The takeaway: Traffic and authority are incomplete proxies for the kind of relevance AI systems increasingly depend on.
How smaller publishers drive big brand influence
When pitching brand content, we typically build a list of journalists whose beats closely align with each campaign’s key insights, which often span several verticals. We then identify the publisher best suited for an exclusive by prioritizing domain authority.
This approach creates a built-in syndication effect, with regional and niche journalists often picking up stories from larger publications. That’s how we regularly earn brand features like this, without ever pitching TV news anchors.
While this remains an effective approach for scaling earned media that builds brand authority and awareness, SparkToro’s data revealed that more vertically focused publishers with minimal traffic (often 5,000-10,000 monthly visits) and mid-tier domain authority (typically in the 60s and 70s) often have much higher audience affinity.
This finding highlights the value of pitching low-traffic, niche publishers that old-school SEO practitioners might once have deprioritized or excluded from outreach.
Across every industry we analyzed, the same pattern emerged: Smaller, niche sites with modest traffic consistently earned the highest affinity scores among the audiences that matter most.
The placements with the greatest strategic value for building entity authority may be the very ones your digital PR team has been overlooking.
Why niche publishers build stronger entity authority than mainstream media alone
I started in SEO in 2006, and the same guiding principle that shaped my career is now being reinforced by a growing body of research: Brand visibility in AI is built through a diverse, authoritative network of brand mentions.
Monthly organic traffic remains a weak proxy for audience alignment, and AI systems synthesizing information about your brand may weigh that distinction more heavily than traditional ranking algorithms do.
We plotted audience affinity against monthly organic traffic, and two distinct clusters immediately emerged.
“Hidden gem” publishers clustered in the upper-left quadrant: high affinity, low traffic. Major publishers occupied the lower-right: high traffic, low affinity.
The 1.7x affinity gap and 130x traffic gap tell the same story from opposite directions: Reach and relevance are inversely correlated more often than most earned media strategies account for.
What stands out in this dataset isn’t just that niche sites outperform on average. It’s how consistently they appear at the very top of the affinity range.
The highest-scoring publishers across industries, including recruitingdaily.com (93), clubindustry.com (90), and chimecentral.org (86), attract just 2,000-10,000 monthly visits.
Meanwhile, many of the largest, most recognizable publishers fall in the 50–65 affinity range, with some scoring in the teens despite attracting hundreds of thousands of monthly visits.
Ultimately, the smartest media mix isn’t mainstream or niche. It’s both. Major publishers still matter for scale, authority, and SEO value, but their broad audiences make beat-level relevance even more important.
The more precisely you place a story within the right category on a large publication while also earning coverage from niche, high-affinity outlets, the more effectively you compound brand visibility across SEO and GEO.
The best media strategies engineer both reach and relevance.
YouTube and Reddit dominate platform affinity
High-affinity authority isn’t limited to publishers.
When we expanded the analysis beyond traditional editorial outlets, YouTube, Reddit, and podcasts repeatedly emerged as influential audience hubs. That matters because AI-driven discovery is increasingly shaped by the broader web of brand mentions.
This shift should expand the definition of earned media. A founder interview on a niche YouTube channel, a data-led discussion in a relevant subreddit, or a subject-matter expert appearance on an industry podcast can all reinforce the same entity associations as a traditional article.
The format may differ, but the strategic value is the same: repeated third-party validation around the topics you want your brand to be known for.
The mistake is treating these channels as post-publication promotion. They should be part of your media strategy from the beginning.
If your campaign is built around original research, expert commentary, or proprietary data, plan how that story can be repurposed to earn brand mentions across publishers, YouTube, Reddit, podcasts, newsletters, and social communities before the first pitch goes out.
Transforming your digital PR strategy to drive brand visibility in AI
I entered SEO when authority could be manufactured through paid link networks. Today, AI systems infer authority very differently by recognizing patterns of trusted, third-party validation across the web.
That shift changes how you should think about earned media. Success isn’t measured only by the authority or traffic of the publications that mention your brand. It’s also shaped by whether those publications reach the audiences you care about and reinforce the topics you want AI systems to associate with your brand.
Traffic and domain authority still matter, but they don’t tell you whether your brand is being reinforced across the sources AI systems retrieve from. Audience affinity adds another layer, helping you identify the placements most likely to strengthen your visibility across AI search experiences.
The takeaway isn’t to stop pitching top-tier press. It’s to stop treating top-tier press as the whole plan.
A GEO-ready media list isn’t bigger. It’s better balanced.
Start with authoritative mainstream publishers. They still matter for reach, trust, link equity, and broad brand validation.
Add high-affinity niche publishers. These outlets may have smaller audiences, but they often concentrate the exact buyers, practitioners, or decision-makers you want to reach.
Include community-driven platforms. YouTube channels, podcasts, Reddit communities, newsletters, and other trusted communities can reinforce the same entity associations as traditional editorial coverage.
Score opportunities by entity relevance. Ask whether each placement connects your brand with the topics, competitors, use cases, and customer problems you want AI systems to associate with you.
Measure more than links. Track high-affinity placements, branded co-occurrences, AI citations, AI mentions, and recommendation-style visibility across major AI platforms.
AI visibility compounds through repetition. A well-placed story on a high-affinity publisher can reinforce the same entity associations as a mainstream feature, especially when it’s republished, cited, discussed by subject matter experts, and adapted across channels.
That’s how a single earned media campaign can shape more than rankings. It can influence the broader context AI systems use to retrieve, describe, and recommend your brand.
The brands that win in AI search won’t necessarily be the ones with the most optimized websites. They’ll be the ones most consistently validated across the sources their audiences and AI systems trust.
Ten years ago, a negative piece of online content primarily affected search rankings.
Today, that same article can influence across Google’s AI Overviews and other AI search experiences. It can be summarized, cited, and redistributed, making it more influential and longer-lasting than it ever should be.
As a result, outdated stories can resurface long after they disappear from traditional search results. That gives older content renewed visibility and makes reputation management far more difficult.
When old articles resurface
I recently saw this happen with a client who owns a grocery chain in the Midwest that has grown successfully for more than two decades.
In the mid-2010s, one location received negative press over a customer service issue. The problem was resolved shortly afterward, and the article gradually faded from public attention.
Years later, AI Overviews gave the story new visibility. Seemingly overnight, the article became a recurring source in AI-generated answers about the business.
A single, outdated news story began shaping how AI systems described a company whose reputation had long since moved on.
AI search engines don’t just retrieve information. They generate answers by relying on published sources they consider reliable.
That changes the role of negative news articles. Even if an article no longer ranks prominently in traditional search results, it can remain an authoritative source for AI-generated answers.
Media coverage often carries strong authority signals. If a negative article receives attention, citations, or discussion, AI systems may continue treating it as a reliable source long after the underlying issue is resolved.
That’s why a single article can influence how AI describes a person, company, or brand years after it was published. The article doesn’t need to dominate search rankings anymore. It only needs to remain a trusted source.
Five or 10 years ago, handling negative content involved suppression. We aimed to bury the negative content by publishing fresher, more positive, and more accurate content, optimizing a client’s online profiles and social media, and building microsites to strengthen its reputation.
That approach matters less today. AI systems readily access and cite original negative sources, even when those sources no longer rank prominently in search results.
AI has changed online reputation management, but you still have options. Here are the approaches we’ve found most effective.
Diversify your sources
To combat those negative news articles, it’s imperative to build new sources that are credible and that present across multiple trusted platforms. The aim is to publish articles on respected outlets, focusing on thought leadership and expert insights — with factual resources to boot.
Respond faster and smarter
Be proactive rather than reactive. Before a negative news source becomes widely cited, get on top of it. Address it with responses that clarify the initial controversy.
Build content that’s citation-worthy
Perhaps the best way to counter an original negative news source is to trump it with citation-worthy content.
Remember the grocery chain I mentioned earlier? To thwart the original negative news article, we focused on publishing original case studies and expert insights tied to the success of the grocery chains. We made sure these pieces were all published on reputable, longstanding media outlets.
Monitor visibility on AI platforms
Returning to the topic of being proactive, the best way to do so is by constantly monitoring your brand.
It’s not enough anymore to see how your brand appears on everyday search engines. You must track how you appear in Google AI Overview and other generative search tools. Spend a few hours every month or so typing queries about your brand into various AI search engines.
Several tools can help you detect negative narratives earlier and monitor how AI platforms present your brand.
Tools like removenews.ai simplify outreach to publishers.
Paste an article URL, and removenews.ai generates a personalized removal request and identifies the editor’s contact information, making it easier to request updates or removal. The tool is free and takes about a minute.
Monitor AI visibility and citations
Need to understand how AI platforms describe your brand? Tools such as Otterly.ai, Mangools, and Ahrefs Brand Radar can monitor citations, visibility, and sentiment across AI search experiences.
Continue using traditional ORM tools
Don’t abandon your existing ORM and digital PR tools.
Platforms such as Semrush and Surfer continue to expand their capabilities, making them valuable additions to an AI-focused reputation strategy.
My 8-year-old daughter desperately wanted a Nintendo Switch. Her evil parents refused to buy it for her.
She was too young to get a job, so she did what any resourceful kid would do: she set up a lemonade stand in front of our house.
But she didn’t just put out a table and a pitcher. She ran a high-stakes A/B test.
Her hypothesis was simple: if she could get more people to stop, she could sell more lemonade and buy her Nintendo Switch faster.
Variant A was her two-year-old sister, Julie, stationed out front to attract attention.
Variant B was our dog, Ginger.
I know what you’re thinking.
The dog. Obviously, the dog.
But her sister won. It wasn’t close.
The only metric that mattered
Actually, my daughter didn’t care about the outcome of the A/B test. She didn’t care how many people stopped by the stand.
She cared about one thing, and one thing only:
Did she make enough money to buy the Nintendo Switch?
Marketers have a similar problem right now.
Generative engine optimization (GEO) is the practice of increasing your brand’s visibility in AI-generated answers from platforms like ChatGPT, Gemini, Perplexity, and AI Overviews.
We’re tracking AI visibility, citation share, impressions, rankings, and every other signal we can find.
Meanwhile, leadership is asking a much simpler question:
Is any of this helping the business grow?
I use a simple test I call the Dollar Rule: If I can’t put a dollar sign in front of a metric, it’s a channel metric, not a business metric.
That’s the challenge with GEO.
Most of the metrics we’re tracking are useful operational signals. They tell us what’s happening inside the channel.
Leadership wants something different.
They want to understand business impact.
GEO arrived at exactly the moment attribution started becoming less reliable.
Traditional SEO measurement was built around a straightforward model: someone searched, clicked, visited your website, and converted. You could trace the path and measure the outcome.
AI search changed that.
Buyers are making decisions before they ever reach your website and AI influence is hard to measure with traditional attribution models.
AI search broke attribution
Buyers now discover brands through AI-generated answers, citations, publishers, forums, reviews, videos, and other sources that influence decisions before a click ever happens. Much of that influence never shows up cleanly in analytics.
That’s why so many teams are struggling to justify GEO investments. The visibility is real. The influence is real. But the attribution is often incomplete.
Waiting for perfect attribution is becoming a convenient excuse for inaction.
If you want buy-in for GEO, you need a way to connect that influence to business outcomes, even when you can’t connect every interaction to a conversion.
Making the case for GEO using financial impact
The biggest mistake marketers are making right now is trying to prove attribution before proving value.
Before you worry about attribution, ask whether you’re measuring something that matters to the business.
That’s where the Dollar Rule comes in.
We’ve found that justifying GEO usually comes down to three things:
Align metrics to business outcomes.
Verify that the metrics reliably point you in the right direction.
Translate the metrics into language your CFO understands.
The Dollar Rule is simple:
If a number doesn’t translate into dollars, it’s a channel metric, not a business metric.
Consider revenue opportunity, revenue at risk, payback period, and customer acquisition cost. These are the metrics that live on a P&L, and they’re the ones your leadership team actually cares about.
CFOs don’t allocate budget based on attribution models. They allocate budget based on expected financial outcomes.
Here’s what that looks like in practice.
Influence over attribution
AI search didn’t just change discovery. It changed measurement.
Traditional organic attribution assumes a simple path: search, click, visit, convert.
AI platforms increasingly answer questions before a click happens, influence buyers across multiple touchpoints, and often remove the referral data marketers depended on.
The result is a strange situation: your GEO campaigns may be influencing pipeline while your analytics platform struggles to prove it.
Loamly estimates that roughly 70% of AI-influenced traffic appears as Direct traffic in GA4, making a large portion of AI’s contribution difficult to trace through traditional attribution models.
That doesn’t mean measurement is impossible. It means we need to broaden where we look for evidence.
Instead of asking, “How many clicks do we get from AI search?” ask:
Is branded search growing?
Are prospects arriving already familiar with our positioning?
Are we cited in AI answers for revenue-driving questions?
None of these signals is definitive on its own. Together, they create enough confidence to make investment decisions.
This is how GEO measurement differs from traditional SEO. You’re not measuring a click path. You’re measuring market influence.
The marketers who adapt fastest will stop treating attribution as a traffic sorting exercise and start combining quantitative signals with qualitative evidence. The goal isn’t certainty. The goal is confidence that your GEO investment is moving the business in the right direction.
You’re measuring the wrong thing
The problem isn’t that SEO or GEO metrics are wrong. The problem is that they’re often precise without being relevant to the business outcome you’re trying to influence. They tell you exactly what happened in a channel, but not whether the business is moving in the right direction.
SEO tools are full of precise numbers. The challenge is that many of those numbers aren’t closely connected to business outcomes.
Precise = exact
Accurate = connected to business outcomes
Leadership would rather have a roughly correct estimate of revenue impact than a perfectly precise count of clicks.
I studied engineering in school. We spent a lot of time talking about precision, as in, how exact and repeatable your measurements are, down to the decimal point. In marketing, those precise metrics look like organic clicks, rankings, impressions, and click-through rate. You can get extremely precise numbers from tools like Google Search Console.
The problem is they aren’t accurate. Accurate measurements tell you whether you’re moving closer to a business outcome that matters. Even if they’re not precise, accurate measurements are more useful because they point you toward the bullseye: business outcomes your leadership cares about.
Knowing you got 40 organic clicks to a page is precise. It tells you almost nothing about whether you’re winning or losing in the market, or in my daughter’s case, whether she’s getting close to buying that Nintendo Switch.
That’s a practical application of the Dollar Rule. When attribution is incomplete, translate the evidence you do have into business impact.
Revenue beats attribution
A rough number tied to revenue beats an exact number tied to channel metrics every time.
When accurate attribution isn’t available, build your case from signals you can actually get your hands on and do the math from there.
Fuzzy math doesn’t replace SEO metrics or attribution. It sits alongside them when a traffic-based attribution metric isn’t available.
Here’s an example:
One of our healthcare clients had a problem.
Prospects were showing up to sales calls already convinced of things that weren’t true.
The source was a competitor’s comparison page that was shaping buyer perceptions long before our client had a chance to tell their side of the story.
We recommended publishing content to counter the narrative, but leadership wasn’t convinced there was enough evidence to respond. So we had to make the case.
SEO tools estimated roughly 40 organic visits per month. Whether that number was right or wrong didn’t matter. It wasn’t measuring influence.
So we looked at something more meaningful.
We talked to our client’s salespeople. They told us that roughly 10% of their qualified B2B discovery calls included unprompted mentions of specific claims from the competitor’s page.
It wasn’t a clean number we could do exact math with, but we couldn’t ignore it. It was real. It was happening on live sales calls.
So we did fuzzy math:
10% mention rate on discovery calls
× 1,200 qualified B2B sales calls per year
× $500,000 average contract value
× 20% average win rate
= $12 million in annualized revenue being influenced by the competitor’s narrative
This wasn’t a forecast, and it wasn’t an attribution model. It was a directional estimate of the amount of pipeline influenced by the competitor’s messaging.
We stopped talking about 40 clicks a month and started talking about $12 million in influenced pipeline.
That’s the number we brought to leadership. Not impressions or citation shares. We brought them twelve million dollars of pipeline being influenced by a page our client was refusing to counter. That is a number a CFO understands.
Lead with the value metrics
If you walk into a GEO campaign review and lead with citation share going up or impressions growing, your CMO is going to yawn. Your CFO is going to wonder what language you’re speaking. In the worst case, they’re going to cut your budget because they don’t see the return.
Here’s how we framed the situation for our client’s leadership:
Leadership funds marketing campaigns with business impact. Translating the problem into dollars changes the conversation.
The decision makers didn’t need certainty. They needed a credible story: leading indicators and momentum that build trust, all tied to dollars.
Focus on what matters
That’s what my eight-year-old intuitively understood at the lemonade stand. Her goal was never to count lemonade stand visitors. Her goal was to buy the Nintendo Switch.
GEO has created a lot of anxiety because it broke the attribution models we relied on for years. But attribution was never the goal.
The real goal: business growth.
If you can connect your GEO efforts to revenue opportunity, revenue at risk, pipeline influence, or customer acquisition, you don’t need perfect certainty to make the case.
You just need evidence that your GEO campaigns are moving the business in the right direction.
Precise metrics tell you what happened. Relevant metrics tell you whether you’re winning.
Before your next GEO report, take every metric on the page and ask one question:
If this metric doubled tomorrow, would the business care?
Then ask the follow-up:
Can I translate this metric into revenue opportunity, revenue at risk, pipeline influence, or customer acquisition cost?
If the answer is no, you’re probably reporting on channel impact, not business impact.
AI shopping is changing what SEO needs to optimize. Structured data, product feeds, entity signals, and crawlable content no longer just influence rankings. They increasingly determine whether AI systems can understand, evaluate, and recommend your products.
The technical foundations haven’t changed. Their role has.
As AI becomes another path to product discovery and purchasing, brands need to strengthen the information AI relies on to make decisions.
AI shopping requires a broader view of brand knowledge infrastructure
For ecommerce and service brands, brand knowledge infrastructure has historically meant maintaining a Google Business Profile, keeping NAP data consistent, and ensuring core pages are crawlable.
Those fundamentals still matter, but they’re now the floor, not the ceiling. Today, brand knowledge infrastructure has three layers.
The static layer
Structured, agent-facing content, including clear return policies, shipping terms, and product differentiation in machine-readable formats. This information needs to be available in crawlable HTML, not hidden behind JavaScript or buried in PDFs.
Agents evaluating whether to recommend your business for a booking or purchase will look for this information the same way a person would check your FAQ page. The difference is they’ll stop looking the moment they can’t parse it.
The real-time layer
Live product and inventory data that AI systems rely on for pricing, availability, and recommendations.
Once a product is added, Universal Cart works in the background to monitor price drops, surface price history, and alert users when an item is back in stock, all powered by Gemini models.
Agents pulling from this system need product data that’s accurate, up to date, and complete at the attribute level. A product listing with a missing shipping estimate or stale inventory count is unhelpful and untrustworthy to the machine making the recommendation.
The entity layer
The signals that establish your brand as a trusted, machine-readable entity across the web. That includes:
Consistent brand naming.
A verified Google Business Profile.
Organization schema with sameAs attributes pointing to authoritative sources.
Accurate Knowledge Graph data.
The entity markup that establishes your organization in Google’s Knowledge Graph is the highest-leverage schema implementation available in 2026. Its impact on AI Mode citations and Knowledge Panel accuracy is substantial and measurable, even though it doesn’t generate visible SERP features.
Traditional SEO asks whether people will click. AI shopping expands that to ask whether machines will trust your data enough to evaluate and recommend your products. These six priorities are where that trust is built or lost.
1. Product data quality
Complete, accurate, real-time product attributes, including titles, descriptions, pricing, inventory, and shipping information, are what AI systems evaluate first. The minimum data set for AI-ready product data includes:
A title.
Description.
Price.
Availability.
Global Trade Item Number (GTIN) or Manufacturer Part Number (MPN).
Shipping speed and cost.
Return policy.
High-quality images.
Stale or incomplete data creates a poor user experience and can prevent your products from appearing in AI-generated comparisons and recommendations before a person ever has a chance to see them.
Audit your product feeds the way you audit technical SEO: systematically, on a regular cadence, and with the assumption that every gap has a cost.
Prioritize price and inventory accuracy first because those are the attributes AI systems verify most aggressively against real-time signals.
2. Machine-readable product information
JSON-LD Product markup, availability signals, pricing data, and shipping details make up the machine-readable layer AI systems parse before anything else.
Implementation best practices haven’t fundamentally changed, but validation requirements have expanded to include AI Mode considerations that existing tools don’t directly measure.
The current validation workflow requires two checks: Google’s Rich Results Test for traditional eligibility and a manual review of AI Mode citation behavior for your key queries.
Beyond Product schema, one of the most underused implementations is Organization schema with knowsAbout and sameAs properties. These establish your entity identity in Google’s Knowledge Graph and improve your chances of being selected as a cited source in AI Mode responses.
3. Structured content beyond schema
Schema markup tells AI systems what your data is. Structured content determines how that data is presented on the page. AI systems evaluate both independently.
In practice, this means three things:
Product specifications should appear in HTML tables, not prose paragraphs. An AI system assembling a comparison interface needs clean, scannable attribute rows, such as material, dimensions, compatibility, and weight, not a sentence that happens to contain those facts.
Policies that influence purchase decisions, including returns, shipping terms, and warranties, should be hosted in crawlable HTML at a stable, linkable URL, not in a JavaScript accordion, modal, or PDF.
If you publish comparison content, such as “our product vs. competitors,” present it as tabular data. AI systems building real-time product comparisons can extract information from structured tables more reliably than from narrative copy making the same claims.
This is as much a content production and CMS decision as it is an SEO one, and it’s worth auditing separately from your schema implementation.
With Google’s Universal Cart and generative UI both pulling from live product data, the quality of your real-time feeds is no longer just a commerce operations problem. It’s an SEO problem. Feeds that update infrequently, omit key attributes, or contain stale inventory signals will underperform in AI-generated shopping experiences, much like slow page speed underperforms in traditional search.
If you use a feed management platform, audit the refresh rate and attribute completeness of your Google Merchant Center data. If you manage feeds manually, establish a regular QA process at the SKU level, not just the category level. AI systems building comparison tables or product simulations from live data will skip products they can’t fully populate.
5. AI-ready business information
For service businesses, such as home repair, beauty, and pet care, prepare for the possibility that Google’s AI will call your business on a customer’s behalf.
That means your Google Business Profile services, hours, and pricing need to be accurate, complete, and consistent with what’s on your website.
Your phone staff also need to be ready to answer agent-style queries: specific, structured, criteria-driven questions about availability, pricing, and service scope.
Assume the AI system will check three things before deciding whether to call your business or move on to a competitor:
Your Google Business Profile services list.
Your website’s pricing and availability information.
Your reviews.
If any of these are incomplete or inconsistent, you risk being bypassed without ever knowing it.
6. CRM and transactional data
Consistent brand naming, structured product identifiers in transactional emails, and clean order confirmation data are signals AI systems can use to connect a user’s history to a current purchase decision.
Audit your transactional email stack with this question: If Google’s AI reviewed every order confirmation your brand has sent, could it accurately identify your products, pricing history, and brand identity? If not, those inconsistencies are creating friction in a recommendation process you can’t see.
The organic window is open, but it won’t stay that way
AI shopping doesn’t replace traditional SEO. It changes what successful SEO looks like. The same technical foundations you’ve relied on for years, including structured data, product feeds, entity signals, and crawlable content, now do more than improve visibility. They help AI systems understand your business well enough to recommend it.
Historically, incomplete or inconsistent data might have meant lower rankings or fewer rich results. In AI shopping, it can mean your products never make it into the comparison, recommendation, or transaction in the first place.
That’s why the six priorities in this article aren’t new SEO tactics. They’re established best practices that now carry greater weight as AI becomes another way people discover and buy products.
Brands that strengthen their brand knowledge infrastructure now will be better positioned as AI shopping matures and competition for visibility inevitably increases.
The gap between who you are and who the machine thinks you are has always been an issue in search. After all, this gap is an alignment problem before it’s an AI problem, per se. AI has finally made it legible.
For example, I recently asked four AI engines to explain who a specific company was in plain language. Guess what? The results were as if I’d asked about four different companies. Same business, four identities, and none of them quite fit the bill based on what I knew to be true.
That gap is the whole problem, and it opens long before any AI is involved. SEO runs on a quiet assumption that four things line up:
What your business says it is.
What the search engine decides your business is.
What AI engines cite your business for.
Who your actual buyers are.
We steer by the ranking and trust the rest to follow. Yet they almost never line up, and the gap tends to sit open for years before anyone names it.
Where does this gap come from?
Every technical decision is a signal: the homepage copy, the internal links, the schema, and the brand saying one thing on LinkedIn and another in the sales deck. When these things disagree, they turn into noise that accumulates.
Those decisions get made in different rooms by different teams, including product, brand, content, and sales, which is one reason SEO can no longer work in a silo. The signals it has to reconcile were never SEO’s to set alone.
None of this began with the advent of AI. It reads the same signals Google always has. The only thing that’s changed is its output.
Traditional Google SERPs returned a position in a list you still had to translate, where contradictory signals could sit buried at the bottom of a page nobody scrolled to.
AI instead returns a plain-text paragraph in the first answer a buyer sees. When it detects noise, it either misinterprets your data or ignores it altogether.
That first impression carries more weight than ever because fewer links get shown and fewer get clicked. Take, for example, a randomized field experiment run in early 2026 by researchers at the ISB Institute of Data Science. They found that when an AI summary appears, outbound clicks to publishers fall by 38%. Users don’t feel they’re missing anything. (It’s a working paper, not yet peer-reviewed, so hold it loosely. Still, the older correlational Pew numbers point the same way.)
The Tow Center puts misattributed citations above six in 10, and the button that used to let users correct the engine has been removed. So whatever the AI engine has decided you are, right or wrong, tends to stand.
Figure 1: The three symptoms, from the newest and most visible (outer) to the oldest and deepest (inner). Each with its test.
These are patterns, not a framework. The name matters less than the test behind each one, and each test is something you can run on Monday.
Entity dissonance
When there’s entity dissonance, the engines are misclassifying the business itself: perhaps the wrong category, the wrong location, the wrong founder, or sometimes even a different company entirely.
It’s the most literal of the three, and the oldest issue SEOs have dealt with. This is ground Dixon Jones and Jason Barnard have covered for years: how to get a machine to hold one clear idea of who your brand is.
How to spot it
Ask each engine plainly who your company is. Search your brand in Google and read the knowledge panel, if there is one. What does it anchor to: the product, the free tools, or the blog? Where do the sitelinks and “People also search for” point?
Then pose the same question to ChatGPT, Gemini, and Perplexity, and line the answers up on four axes: category, location, founder, and what it sells.
You can tell there’s entity dissonance when the engines contradict each other, when one fastens the brand to a same-named stranger, when the category is the traffic magnet rather than the product, or when the location is the registered address instead of the market served. The wider the disagreement, the deeper the problem sits in the entity layer.
Audience mismatch
Audience mismatch happens when the traffic a site earns is not the buyers it needs, and the people searching are a different population from the people buying.
In SEO, we’ve called this user intent for years, but it runs deeper than the intent behind any single search. It’s whether the audience you sell to actually needs the product, with everything that implies and every team that has a say in it.
The instinct is to open Search Console, hunt for low click-through rates, and treat it as a keyword problem. It isn’t one. It starts with knowing the buyer: interviews, real voice of customer, and personas built from evidence rather than a demographic sketch.
The concrete version is to set the queries and pages that bring traffic, besides who actually closes in your customer relationship management (CRM) platform, tagged by source and intent, and ask yourself whether the two describe the same person.
You can even let a model stand in for that buyer. Feed it what you know about your target audience, the job they’re doing, their constraints, and the words they use, then have it read your site as them.
Stanford’s research on simulating human behavior with AI agents found that an agent grounded in a two-hour interview with a real person reproduces that person’s survey answers about 85% as accurately as the person does themselves when retaking the same survey two weeks later.
A persona built only from what you know about your ideal customer profile (ICP) is a weaker version of this, but it’s still a useful starting point. It flatters and smooths over the friction real buyers feel, so use it to explore, not settle, the question.
However you run it, your audience is broken when the traffic sits in discovery questions and free tools while the closed-won business clusters around bottom-of-funnel intents like compliance and migration that barely surface in the traffic.
It’s also the pattern where SEO gains the most from leaving its own lane because the people who can tell you who the buyer really is sit on the copy, brand, and analytics teams. If you spend too long inside the algorithm, you can lose sight of the person it’s meant to reach.
Citation drift
Citation drift is when AI platforms do cite the brand, but for things or services it doesn’t sell, such as old content, abandoned free tools, or the reputation it’s trying to outgrow. It’s the newest of the three, and that isn’t a coincidence.
That’s because audience mismatch and entity dissonance have accumulated quietly for years, and citation drift is what surfaced once AI started reading that accumulation back to us in plain text.
How to spot it
Ask each engine what the brand is known for and what it does best, and write down the assets, pages, and topics it names. Beside that list, make a new list of all the products and offerings that actually pay the bills, and rank them by revenue. The distance between them is the drift.
You know citation drift is a problem when the engines praise you for things like free calculators and old blog posts while your paid product goes unmentioned. Measure it as a pattern, not a snapshot. If you ask the same question on different days, the list often looks different, so rerun it before you trust the gap.
The four signals rarely get read against each other, and almost never against what buyers say on sales calls. That last reading never comes off a SERP.
The identity gap audit: An example of one business, four signals
The four identity gap signals I opened this article with were one signal of four. Read the same business through all of them, and the three symptoms surface together in one company, at once.
The audit is real and anonymized. I’ve rounded the figures but kept the proportions exactly as I measured them.
Figure 2: One company, asked who it is, gets a different answer from every signal; the AI alone splits into three. Anonymized client audit.
This business sells accounting software to freelancers and small companies. What brings people to the site is a set of free fiscal calculators (VAT, withholding tax, payroll, and an invoice generator).
What pays the bills is a subscription that keeps those same small businesses’ books in order. Hold that split in mind because it’s where the noise starts. The thing that earns the traffic is not the same as the thing that earns the revenue, and every system in the chain reads the business through its traffic.
What the business says it is
Start with what the company is trying to become. Our example business grew up as one narrow product, a free tool that handled a single fiscal chore for freelancers, and it outgrew that.
Today, it wants to be a compliance platform that small companies trust with their books, judged against accountants and established software rather than other free calculators.
Its own positioning document says exactly that, then admits the catch. What the brand still transmits — the visual language, the channels it grew up on, the words it uses, the entities it gets associated with, and so on — all lag a step or two behind what the business has become. This is a company that already knows it’s being read as something it no longer is.
What the search engine thinks the business is
Most audits stop here, so the gap is easy to miss. This software brand has a knowledge panel, so Google knows it exists. But look at what the panel anchors to.
To the search engine, the site appears to be a free resource and a blog. That’s because the sitelinks lead with the calculators, not the product. The entity is registered to a single address in one country, while the market it serves is in another.
The “People also search for” rail for this company surfaces complaint and legitimacy queries, the quiet version of someone asking whether the company is for real. Google hasn’t filed the business under the wrong heading, exactly. It has filed it by its traffic magnet rather than by what it sells.
Figure 3: Traffic vs. leads, by content type. Anonymized client audit.
What the AI cites the business for
This is the lens the opening came from. Those four AI engines, when asked the same plain question, disagreed completely.
One didn’t recognize the company at all and answered with the generic meaning of the business name.
A second got the founder’s name right, then attached it to a same-named person from an unrelated field. Note that this is not a hallucination but a reconciliation error: two people with one name collapsed into a single identity.
A third engine recognized the company but described it through its old content and its free tools, never through what it charges for.
Only the address-pinned version came close, and it had the geography wrong.
Four machines, four identities, none of them what the company says it is.
Who actually buys from this business
The one signal that none of the machines are reading is perhaps the most important: the buyer. And “the buyer” is really three people:
The audience pulled in by the free tools.
The customer who buys today.
The upmarket customer the business is growing toward, from sole traders to small companies that need real compliance.
The sales calls reveal who actually closes. Across more than 1,300 calls (895 captured a reason the buyer gave for choosing), the intent that wins by a wide margin, close to a quarter of the time, is compliance.
The buyers are essentially asking the business, “Keep me out of trouble in an audit.” Price sits near the bottom of the reasons people give, and the objection that kills the most deals is data migration, the fear that moving the books across will be slow or costly.
So the mismatch hits twice. The current buyers’ real questions, migration and audit risk, go mostly unanswered on the site. And the upmarket buyer never sees anything built for them because none of that shows up in how Google files the business, in what AI cites, or in the calculators that bring the traffic.
So the four signals each answer “Who is this?” differently, and the buyer’s answer, the one that decides the sale, is the one none of the machines can read. Read back through the lenses, and all three symptoms are there at once.
The search engine and the AI engines misclassify the entity. AI cites the free tools instead of the product. The buyer asks for something none of the traffic reflects. One cause sits underneath all three: the traffic magnet pulling the brand’s identity away from what it sells. The rest of the work is closing that gap.
The work closing the gap that got skipped
Closing the gap is two jobs, not one. The first is the SEO everyone already does. The second is the part that gets skipped, and it’s where the deals actually live.
Find the gaps the tools miss
Most of the SEO here is the SEO everyone does, and it’s necessary work: keyword research by topic, competition, and trend that produces a list of terms with volume and difficulty.
Doing only this kind of SEO skips the more difficult component. You have to map the business against the buyer’s actual journey, every doubt from first look to ready to pay, and make sure something on the site answers each one.
Map that against real sales calls, and you’ll likely find that the holes aren’t where the keyword tool says they are. In this audit, the questions that closed deals — “Can I migrate last year’s books?” “Am I covered if I’m audited?” “What happens to my data?” — barely registered as keywords.
The volume instead sat at the top of the funnel — “How to write an invoice” and “VAT calculator” — the things people search before they care who you are. The decisions got made on questions the tool couldn’t see.
Be precise here because it’s a claim a fact-checker should push on. Zero measured volume doesn’t mean nobody asks. It means the buyer’s own phrasing falls below the tool’s floor, and a closing question, asked once at the bottom of the funnel and phrased a hundred ways, doesn’t aggregate the way a discovery term does. The questions that close a deal live below the line the keyword tool can see.
That zero-volume queries can still matter isn’t news. SEOs have made that case for years. What’s new is that the engines now run on them. Query fan-out, the way a model spins one prompt into subqueries that Mike King and Dan Petrovic have each mapped closely, lives almost entirely in that blind spot.
The funnel map is only half of it, and the smaller half. The bigger job is cleanup. You have to:
Fix the entity dissonance so the engines stop confusing the company with a calculator site and a same-named stranger.
Close the topic gaps where the buyer’s real questions went unanswered.
Open the niche outward toward the upmarket buyer the catalog never spoke to.
When you’ve done this, it’s time to prune content. You thin out the free content and the generic explainers dragging your brand’s whole identity toward the traffic magnet and away from what it sells.
That pruning is the part that feels backward, yet matters most. Accept losing some traffic on purpose because the traffic was noisy. Clean the signals, and two things happen together:
The engines start to recognize you for what you actually are.
Your real buyer starts to find you.
Those turn out to be two sides of the same coin. When you close the distance between who you are and who the machine reads you as, you’ve closed it for the buyer, too.
A site reorganized around the buyer’s problem doesn’t just earn more traffic today. It changes what it can earn tomorrow.
AI works by matching a need to an answer, so a site shaped that way gets found twice: once for the search buyers run today and again for the conversation they have tomorrow.
This is an SEO problem, not an AI problem
It’s tempting to read this as a reason to chase the chatbots, to optimize for ChatGPT the way we once optimized for Google. That’s the wrong instinct.
The AI layer didn’t create the mismatch between who you are and who the machine thinks you are. It inherited it from the search layer and removed the user’s ability to correct it. The fix lives upstream, where it always did: the entity layer and your positioning.
Two things make achieving this more difficult than it sounds. What AI says about you doesn’t hold still. SparkToro’s experiment found that asking ChatGPT or Google’s AI for brand recommendations a hundred times returns the same list fewer than one time in a hundred, and the same order roughly one time in a thousand.
You can’t optimize a position that doesn’t survive two identical prompts. You can only make the underlying entity unambiguous enough that you surface more often. And, in a sense, the churn is beside the point.
What sits under it is personalization, every user getting a different answer, and you don’t win that by chasing positions. You win it by speaking clearly to the audience you actually want, the one thing that stays constant across all those different answers.
Ranking no longer guarantees a citation, either, and the numbers look contradictory at first.
An AI reply tends to pull one well-ranked anchor and several lower-ranked sources from fan-out, so ranking still helps. It just stopped being sufficient. (Part of that 18-month shift is likely vendors parsing citations better, not only engine behavior changing.)
What closing the gap costs, and what to do about it
The cost of the four-way mismatch is paid in two currencies. One is demand that never converts, that is, the traffic earned against discovery terms while the buyer’s actual questions go unanswered.
The other is being cited for the wrong things — your old blog posts and free calculators — rather than the product that pays the bills, leaving the thing you sell invisible. Being cited is the brightest part of this, and the one everyone watches now, but it isn’t what sustains the organic channel. It’s the same mismatch as the rest, just the part that catches the light.
Neither cost gets fixed downstream with more content or a cleverer chatbot play. The first move isn’t technical at all. Before anyone touches the entity layer or the content, the business, marketing, product, and sales teams have to agree on who your company is, what it sells, and to whom.
Most organizations never write that down, so the same argument gets refought on every campaign, every page, every release, and each team settles it a little differently. That’s where the noise is born.
A single internal source of truth, the company’s own reference document for who it is and who it serves, is what keeps the four signals from drifting apart again. Without it, you pay for the same decision and the same risk, over and over.
Figure 5: Fix the source of truth and the four signals converge on one answer. The mirror of Figure 2.
The four signals will never line up on their own. The job is to notice when they’ve come apart and close that distance before an answer engine quotes the gap back to a buyer as fact.
OpenAI is discontinuing ChatGPT Atlas, its standalone desktop browser. The browser-based AI features are moving to the new ChatGPT desktop app, which includes ChatGPT Work, OpenAI’s work-focused agent, alongside ChatGPT Codex.
The end of Atlas. James Sun of OpenAI confirmed on X Atlas will be deprecated Aug. 9.
“The current targeted date for deprecation is 8/9, and we’ll share more information in the upcoming days both in-app and via email,” Sun said.
One desktop app. The new ChatGPT desktop app becomes OpenAI’s primary desktop product with built-in browser capabilities. Instead of maintaining a separate AI browser, OpenAI is combining browsing, work-agent features, and Codex into a single app.
Chrome users can keep Chrome. OpenAI also offers a ChatGPT and Codex extension for Chrome. That lets Chrome users access ChatGPT in their existing browser without switching to an OpenAI browser.
Why we care. OpenAI is moving AI browsing from a standalone browser into the main ChatGPT app, where more users can ask questions, research brands, and complete tasks. That gives ChatGPT another way to shape discovery beyond traditional search results.
Google unveiled Gemini Intelligence at the Android Show on May 12, alongside a new laptop called the Googlebook. The company describes Gemini Intelligence as a layer that runs beneath the Android operating system across laptops, phones, watches, and glasses.
The new Googlebook is built from the ground up around an AI agent that understands what’s on your screen and acts on it for you. Point at a date in an email, and it’ll set up a meeting. Select pieces of furniture in an app, and it’ll show you what they’d look like in your living room.
Now that an operating system can complete tasks without users even opening a webpage, how people search, discover, and conduct commerce will fundamentally change. Let’s look at how this will affect the search industry.
What the shift to an agentic operating system means
Up until now, a person had a question or intent, typed it into a search engine, received a list of links, and chose one. Getting your website to rank on that list was the prize, and the entire SEO industry was built around earning that click.
Gemini Intelligence assumes something completely different. A user still has search intent, but an AI agent now handles the middle steps — reading pages, filling out forms, and, increasingly, completing the task for you. Instead of you visiting a website, an AI agent visits it on your behalf.
One example is Chrome Auto Browse, launched in January and built on Gemini 3. It handles multistep tasks like researching flights, filling out forms, scheduling appointments, and managing subscriptions, then pauses to ask before making a purchase.
A 2025 preprint evaluated the declared-tools approach across online shopping, authentication, and content management. It found that handing an agent pre-structured interaction data cut processing requirements by 67.6% and reduced costs by 34% to 63%, compared with parsing the full HTML document. Task success was only slightly lower than with the traditional method: 97.9%, compared with 98.8%.
AI agents prefer sites they can transact with cleanly because it’s more efficient. Gemini Intelligence only works if agents can reliably perform tasks on websites.
Two protocols make this possible: WebMCP makes a site’s actions callable, and the Universal Commerce Protocol (UCP) allows an agent to complete a sale. Together, they let an agent finish the job without a human having to load a page.
WebMCP
This API lets a website declare its functions as structured tools an agent can call, such as searching inventory, starting checkout, or submitting a support request. This effectively lets you hand an AI agent a labeled menu.
Google co-developed WebMCP with Microsoft. An origin trial is live in Chrome 149, Firefox has committed to the third quarter of 2026, and Safari is expected to follow in the fourth quarter.
Universal Commerce Protocol (UCP)
This protocol gives AI agents a common language to discover products, build a cart, complete checkout, and handle orders without a user visiting the site. Google also has a consumer-facing surface layer called Universal Cart, which collects items as you move across Search, Gemini, YouTube, and Gmail.
Google, Shopify, Walmart, Target, Etsy, Wayfair, PayPal, and Stripe co-developed UCP, which launched in January.
Websites are rapidly changing from destinations to backends, from places people visit to places agents quietly use. The operating system is becoming the search layer. The question is no longer whether you rank, but whether an agent can use your site.
To prepare, audit your most valuable actions, whether that’s a lead form, booking flow, or checkout page, and ask whether an agent could complete them instead. Check your Lighthouse Agentic Browsing score the way you check Core Web Vitals to see whether an agent can use your site in addition to reading it.
If you run ecommerce, find out whether your checkout is reachable through UCP or ACP. Keep doing the retrieval work, because an agent still has to find and trust you before it can act on your behalf.
Login
