There is a tendency in SEO, and now even more in what we call AI Search optimisation, to look at the website as if it were the complete universe of a brand.
Make the website technically perfect. Make every important page crawlable and indexable. Build a clean information architecture. Define the ontology of the domain. Organise its taxonomy. Cover the relevant entities and their relationships. Understand the micro-moments surrounding the customer journey. Create the right content for every stage of that journey. Add structured data where it helps machines disambiguate what we mean. Strengthen internal linking. Publish original research. Generate real information gain.
I agree with all of this.
In fact, I have spent quite a lot of time writing about precisely these things.
In The Architecture of Authority, for instance, I argued that successful content strategy should be understood as engineering rather than publishing: we build interconnected information systems designed to answer the questions of a specific audience instead of producing disconnected pages because a keyword appeared in a spreadsheet.
More recently, in The Consensus–Information Gain Axis, I argued that repeating what everybody else already says turns content into a commodity, while pure novelty without a foundation of established knowledge creates a different problem. The interesting position lies between those two extremes: understand the consensus, and then contribute something that meaningfully expands it.
But there is another part of authority that we sometimes underappreciate.
We can engineer almost everything that happens inside our website.
We cannot engineer authority entirely inside our website.
At some point, somebody else must have a reason to talk about us.
And this is where Digital PR, traditional outreach, journalism, collaborations, expert commentary, original research distribution, interviews, podcasts, conferences, and the broader machinery of brand amplification become much more than “off-page SEO”.
They become the other half of the authority system.
Companion resource: a free spreadsheet mapping 202 publications to AI publisher deals, with A/B evidence levels so you can separate what’s documented from what’s inferred.
Brand authority in AI search: Owned vs. Distributed Authority
Let me simplify the problem.
There are two information environments in which a brand exists.
The first is the environment the brand controls. I will call this Owned Authority Architecture.
Your website belongs here, together with all the knowledge and relationships you explicitly construct inside it.
Imagine that we work for a cybersecurity company. On our own website, we can establish that the brand offers ransomware protection, serves financial institutions and develops specific products; that a particular person is its CEO; that another person authored its research; and that this research analyses ransomware trends.
We can reinforce those relationships through navigation, internal links, contextual anchors, structured data, author pages, product architecture, and countless other mechanisms.
We can also construct a comprehensive representation of the subject itself.
This is why ontology matters.
An ontology defines what exists in the domain and how the concepts relate to each other. A taxonomy helps organise those concepts into useful classifications. Context tells humans and machines what those things mean in a specific situation.
Add buyer journeys and micro-moments and the architecture becomes even richer because we are not simply modelling a subject. We are modelling the questions people have while moving through it.
This work is essential.
However, there is an unavoidable limitation: most of those assertions still originate from us.
We are essentially telling the world:
Trust us. We understand cybersecurity.
We may provide extraordinary evidence for that claim. We may publish research, demonstrate expertise, and produce the best technical resource in the industry. Yet, as long as the evidence primarily lives inside properties we control, part of that authority remains self-declared.
The second environment is what I call Distributed Authority Architecture.
It consists of the network of independent documents, publishers, journalists, researchers, creators, institutions, and communities in which our brand and its people appear.
Suppose our CEO is repeatedly consulted by journalists whenever an important ransomware attack occurs. One journalist quotes her about ransomware economics. Another asks her to comment on the vulnerability exploited during an attack. An industry publication interviews her about regulation. A conference invites her to discuss threat intelligence. A university report cites research produced by the company.
Now the relationship between the brand and the subject no longer exists only on the brand's website.
Across independent documents, the same associations start appearing again and again: this person leads this company; this person is consulted about ransomware; this company produces research about the subject; third parties consider that research or expertise worth referencing.
This is what I mean by distributed semantic corroboration.
I am not suggesting that Google has a ranking factor called “Distributed Semantic Corroboration”, nor that ChatGPT calculates some secret Brand Authority Score every time a CEO appears in a newspaper.
That would simply be inventing another SEO mythology.
I am describing the information environment that we can observe.
If many independent documents associate the same entity with the same domain of expertise, the web contains substantially more evidence of that relationship than if the association exists only on the entity's own website.
That distinction matters to humans and journalists, certainly. It also creates richer evidence for search systems and retrieval-based AI systems trying to identify relevant entities, sources, and relationships.
Brand amplification through digital PR: Distribute associations, not just URLs
This is where I think we need to reconsider how we talk about Digital PR.
Historically, SEOs often looked at PR through a very narrow window.
The objective was the link.
Did the journalist link? Is it followed? What is the authority of the domain? How much organic traffic does it have? Can we somehow convince them to transform that brand mention into an anchor?
Those questions are not useless. Links still matter.
But when we reduce PR to link acquisition, we miss most of what actually happened.
Imagine that the Financial Times interviews the CEO of our fictional cybersecurity company and quotes her explaining why a new kind of ransomware attack is becoming more common.
There may be a link.
There may not be.
But something more interesting than the creation of one edge in a backlink graph has happened.
A respected independent source has created a document in which our company, our CEO and a specific area of expertise coexist within a meaningful context.
The journalist has added another piece of evidence to the information ecosystem.
For this reason, I find it useful to distinguish four different outcomes of media exposure:
A mention is the appearance of our brand or one of its relevant entities.
A citation, in the practical Digital PR sense I use here, is a mention accompanied by a link to the originating or supporting source.
A quotation directly associates a person with an attributable statement.
A source attribution associates information, research or a factual claim with its origin.
From an SEO perspective, we have historically been disproportionately obsessed with the second.
From the perspective of brand amplification, all four can matter.
An interesting Ahrefs analysis of 75,000 brands found relatively strong correlations between branded web mentions and visibility in ChatGPT, AI Mode, and AI Overviews, while traditional authority metrics such as Domain Rating showed substantially weaker correlations. Ahrefs explicitly warns that correlation does not demonstrate causation, and that warning matters, and the study does not prove that acquiring another hundred mentions will mechanically produce more ChatGPT mentions. What it does show is a pattern consistent with the idea that brand prominence across the web cannot be reduced to backlink counts alone.
The strategic shift is subtle but important.
Digital PR doesn't simply distribute URLs. It distributes associations.
And associations are one of the fundamental components of meaning.
The missing bridge between Information Gain and Digital PR
This gives us a missing piece in the Consensus–Information Gain model.
Novel information does not remain novel forever.
Suppose a company publishes genuinely original research showing that 37% of a specific kind of transaction now takes place through a mechanism that nobody had previously quantified.
At the moment of publication, the information gain is high.
If the research is credible and useful, other people begin referring to it.
Eventually, that statistic may become common knowledge within the industry.
Its novelty has decayed. What was once information gain has progressively entered the consensus.
The original source, however, can retain a different advantage: it may continue to be recognised as the originator of the information.
There is an intermediary step here that deserves much more attention.
Information Gain must travel.
We sometimes act as though publishing original research automatically transforms the information landscape.
It doesn't.
You can publish the most fascinating dataset of the year on your corporate blog and discover that almost nobody noticed.
The information exists.
Its distribution does not.
This is where PR enters the system.
A useful representation of the process is:
Information Gain → Distribution → Independent Contextualisation → Propagation → Emerging Consensus
The company creates the original information. PR introduces it to journalists, specialists and other relevant actors. Those independent sources decide whether it deserves attention. Some report it. Some question it. Some contextualise it with other evidence. Some ask the company's experts to explain its significance. Others later discover the story through that first wave of coverage.
Gradually, the finding moves away from being something that exists only on company.com and becomes part of the information environment surrounding the topic.
This is very different from manufacturing mentions because we believe mentions are an AI Search signal.
The causality goes the other way.
The company has produced something useful enough to generate independent interest, and that interest creates more documents in which the brand, its expertise, and its original contribution are placed into context.
In other words, Digital PR can act as the distribution layer of Information Gain.
For years we have said that brands should become publishers.
That was directionally correct, but incomplete.
Brands should also aspire to become sources.
First-generation and second-generation brand amplification
Now we arrive at a phenomenon I find particularly interesting.
Let's imagine that our research earns coverage from the BBC.
The BBC publishes an article mentioning the company and quoting its CEO.
That article is what I call a first-generation mention: the coverage directly generated by our activity.
But information on the web does not behave like a museum object permanently locked inside the page where it was first published.
A specialist industry publication may see the BBC story and develop its own analysis. A regional publication may use the finding because it has local implications. A trade association may add the statistic to its weekly newsletter. A blogger may discover the specialist article and discuss it. A journalist working on another piece two months later may encounter several of those documents and contact the CEO again.
Those are second-generation mentions.
Continue the process, and we eventually reach what we might call Nth-generation amplification.
Conceptually, the propagation may look something like this:
Brand / Original Research → BBC → Industry Publication → Trade Association or Specialist Newsletter → Independent Analysis → Further Journalism and Discussion
At this point, it becomes difficult to describe what happened purely as “link building”.
What we are observing is closer to an information propagation graph.
Academic publishing provides a useful analogy.
The importance of an academic paper isn't defined only by the journal in which it originally appeared. Its influence propagates through papers that cite it, papers that cite those papers, meta-analyses, textbooks, conference presentations, and eventually the accepted knowledge of the discipline.
Journalism and web publishing are obviously not identical systems, but the propagation principle is similar.
Interestingly, Google has previously recognised a related phenomenon in news discovery. When it introduced its “highly cited” treatment for news, Google described exactly the case of an original story being picked up by numerous other outlets, using those references to help readers discover the reporting that contributed distinctive value. Google's explanation of the feature is useful because it shows how explicit citation relationships can help identify an originating source within a wider news ecosystem.
For Digital PR, this means that the value of a placement cannot always be estimated by looking only at the original URL.
Some publications are powerful partly because journalists read them.
Their stories become inputs into other stories.
That gives us another useful concept for PR planning: Propagation Potential.
How likely is coverage in Publication A to generate further coverage, discussion, citations or discovery elsewhere?
This becomes especially interesting when AI accessibility enters the picture.
The new variable: proximity to AI information ecosystems
During the last few years, OpenAI, Google, Perplexity and other AI companies have developed direct relationships with publishers.
These agreements are not all the same.
Some concern content licensing. Others allow summaries or excerpts to appear inside AI products. Some involve direct feeds, APIs, revenue sharing, archives or product collaboration.
Therefore, creating a column called “AI deal: Yes/No” and pretending that every row means exactly the same thing would be analytically poor.
Nevertheless, those relationships give us useful information.
Take OpenAI.
By January 2025, OpenAI described relationships with news organisations representing hundreds of newsrooms and content brands. Its publisher ecosystem included Associated Press, Axel Springer, Condé Nast, Dotdash Meredith, Financial Times, GEDI, Hearst, Le Monde, News Corp, PRISA, Reuters, The Atlantic, TIME and Vox Media, among others. OpenAI's Axios announcement provides a useful snapshot of how broad that media strategy had already become.
Some agreements are remarkably broad.
The OpenAI–Future agreement covers more than 200 media brands, including Marie Claire, PC Gamer, TechRadar, Tom's Guide, The Week, Who What Wear and Cycling Weekly. ChatGPT users can access content from across Future's portfolio with attribution and links to the original articles.
Other partnerships identify individual publications or narrower groups. OpenAI's News Corp relationship, for instance, encompasses major publications across the United States, the United Kingdom and Australia, while its Schibsted agreement specifically covers titles including VG, Aftenposten, Aftonbladet and Svenska Dagbladet. The News Corp partnership and the Schibsted agreement illustrate how different the scope of individual publisher arrangements can be.
Google has taken a somewhat different path.
In December 2025, it announced a commercial AI publisher pilot involving titles including Der Spiegel, El País, Folha de S.Paulo, Infobae, Kompas, The Guardian, The Times of India, The Washington Examiner and The Washington Post. The program includes experiments with AI-powered article overviews and audio briefings in Google News. At the same time, Google identified Associated Press, Estadão, Antara and Yonhap as providers of real-time information used to enhance results in Gemini. Google describes the different arrangements here.
Perplexity built yet another model around publisher participation. Its programs have included TIME, Der Spiegel, Fortune, Entrepreneur, The Texas Tribune and other organisations, while Comet Plus launched with publishers including Condé Nast, Fortune, Le Figaro, Le Monde, Los Angeles Times and The Washington Post.
These relationships are strategically interesting.
But this is exactly where we need to stop before transforming a useful observation into an imaginary ranking factor.
A publisher agreement is not an AI visibility guarantee
Suppose The Washington Post has a commercial relationship with an AI operator.
You obtain coverage there.
What can we legitimately conclude?
We can conclude that a documented relationship exists between the publication and the AI company.
Depending on the arrangement, we may also know that content from that publisher can be delivered, displayed, summarised, retrieved, or otherwise incorporated into a particular AI experience.
What we cannot conclude is:
Because The Washington Post mentioned my brand, ChatGPT will now recommend my brand.
There is no evidence for that shortcut.
Nor should we assume that every document produced by a partner publication receives preferential treatment over content from non-partners.
OpenAI itself makes this distinction clear. Its description of ChatGPT Search explains that the product works with publisher partners but also says that any website or publisher can choose to appear in ChatGPT Search.
So why should we care about publisher agreements?
Because they reduce one category of uncertainty.
With an arbitrary website, we may not know whether an operator has a direct feed, whether licensing restrictions affect use of its content, whether it participates in a specific publisher program, or whether some other delivery arrangement exists.
With an explicit partnership, we know that a defined relationship between publisher and AI operator exists.
That is enough to make AI Ecosystem Proximity a legitimate additional variable in Digital PR planning.
Not the first variable, nor the decisive one, but an additional one.
The BBC test
Whenever I create a framework, I like trying to break it.
The BBC is useful for doing exactly that.
As of September 2026, I cannot find a publicly announced BBC content partnership with OpenAI, Google, Anthropic or Perplexity comparable to the relationships described above.
The BBC has also taken a restrictive position toward unauthorised AI use of its journalism. In June 2025, for instance, it threatened legal action against Perplexity over alleged scraping and reuse of BBC content. The Guardian reported the dispute at the time.
Now imagine that you have completed outstanding original research and can pitch it to two publications.
Publication A has a formal content agreement with an AI operator but is only moderately relevant to your industry.
Publication B is the BBC.
Would we seriously reject BBC coverage because it does not appear in our AI-partnership spreadsheet?
Of course not.
The reason is straightforward: the strategic value of a publication does not begin with its relationship to an AI vendor.
A BBC story can reach customers, influence journalists, trigger branded searches, generate secondary reporting, attract citations and introduce a brand into conversations where it did not previously exist. Its importance derives from its position in the information ecosystem.
AI accessibility can add value to that position.
It does not create it.
The BBC, Google and the limits of what we can observe
The BBC example also exposes another problem in the way we analyse AI accessibility.
We like binary answers.
“Is BBC blocked from AI?”
Unfortunately, the real system is much more layered.
Google provides an especially useful demonstration.
Google-Extended, for instance, allows publishers to control whether content crawled by Google can be used for training future Gemini models and for some grounding uses in Gemini Apps and Vertex AI. Crucially, Google's documentation states that Google-Extended does not affect inclusion or ranking in Google Search. Google documents that distinction explicitly.
Then the UK regulatory environment changed.
On June 3, 2026, the Competition and Markets Authority imposed a conduct requirement designed, among other things, to give publishers effective tools to prevent their content from powering generative AI features in Search and to require clear attribution in AI-generated search results. The CMA announcement explains the requirement.
Google consequently introduced a separate control in Search Console.
Verified site owners can decide whether their links and content may appear in and help ground generative AI Search features such as AI Overviews, AI Mode and generative AI experiences in Discover. The control began as a UK test and, according to Google, was rolled out worldwide on August 31, 2026. Google's announcement of the control explains both its purpose and its scope.
And here is the important part for our BBC example.
That preference lives inside Search Console.
We cannot inspect another publisher's Search Console.
Therefore, even if we analyse BBC's robots directives, we cannot know from outside whether BBC has used the Search Console setting to opt out of Google's generative Search features.
This gives us a methodological rule worth remembering:
Observed crawling ≠ training permission ≠ Search inclusion ≠ retrieval eligibility ≠ commercial licensing.
They are different questions governed by different mechanisms.
Google's documentation makes the differences particularly visible. The new Search Console control can prevent content from appearing as links or being used for grounding in the supported generative Search features. Google-Extended governs different AI uses. noindex governs inclusion in Search more broadly. And technical eligibility is still not a guarantee that Google will crawl, index or serve a particular page. Google's Search generative AI control documentation and its documentation for AI features in Search describe these separate layers.
Some aspects of the system are observable, but others simply are not.
Strategy based on evidence needs to respect that boundary.
What happens when the primary publisher is inaccessible?
Return now to our BBC story.
Suppose, for the sake of argument, that the original BBC article is unavailable to a particular AI retrieval system.
The information does not necessarily disappear from the wider ecosystem.
A specialist publication may report the story. A trade association may discuss the finding. A newsletter may cite the specialist coverage. Another journalist may independently interview the company.
The original information can therefore propagate into domains that are crawlable, indexed or otherwise available to systems that could not access the first article.
This is precisely why the first-generation/second-generation distinction matters.
If Digital PR works well, the initial placement creates opportunities for the brand-topic association to appear elsewhere.
Some downstream sources may have direct agreements with OpenAI or Perplexity. Others may appear in Google's index and be eligible for Search features. Others may simply be accessible to crawlers. Some may eventually appear in future training datasets.
Usually, we cannot know every path the information will follow.
And we do not need to.
A strategy should not depend on speculative assumptions about future training corpora.
The objective is to create information valuable enough to be independently repeated, attributed and discussed.
Stop thinking in terms of “SEO publications” and “AI publications”
This brings us back to an old problem: SEOcentrism.
We SEOs are extraordinarily good at taking an important business activity and reducing it to the part that can be measured inside an SEO tool.
PR becomes backlinks. Brand becomes branded search volume. Content becomes keyword coverage. Authority becomes Domain Rating. AI visibility becomes citation counting.
Each measurement can be useful.
The danger appears when the metric replaces the phenomenon.
A publication isn't valuable because Semrush says it has a lot of organic traffic, nor because OpenAI signed a contract with its parent company.
Its value may instead come from its position inside a specific information network: the journalists who read it, the decision-makers it reaches, the stories other outlets pick up from it, the specialist vocabulary it helps define, or the authority it has accumulated around a particular subject.
Search engines operate inside these networks, AI Search systems do too, and marketing certainly does.
The useful question, therefore, is not “Which publications are good for SEO?” or “Which publications are good for ChatGPT?”
It is:
Which information environments matter for the entities, topics and audiences my brand needs to be associated with?
Once we answer that question, AI accessibility and publisher relationships become another analytical layer.
A practical model for Digital PR prioritisation
If I had to represent the logic in one conceptual formula, I would use something similar to this:
Digital PR Opportunity = Editorial Relevance × Topical Authority × Audience Relevance × Story Fit × Propagation Potential × Search Visibility × AI Ecosystem Proximity
This is not a mathematical model.
Please do not create a spreadsheet tomorrow, assign every factor a number from one to ten, multiply everything together, and announce the creation of the “Fiorelli AI PR Authority Score”.
That would be missing the point spectacularly.
The formula is simply a way of remembering that publication value is multidimensional.
Take a highly specialised cybersecurity publication with no known OpenAI agreement.
Perhaps every CISO reads it. Journalists at larger technology publications monitor it. Its editorial team understands the technical nuances of the industry. Its stories rank well and frequently generate discussion elsewhere.
That publication may be vastly more valuable than a generic lifestyle publication belonging to a media group with an AI partnership.
Now imagine two publications that are equally relevant, equally respected and similarly likely to cover our story.
One has no observable AI relationship.
The other is explicitly part of a publisher agreement that places its content inside one or more AI experiences.
At that point, AI Ecosystem Proximity can sensibly become a tie-breaker or an additional priority signal.
That is the role I see for the database accompanying this article.
Building a publication-level map of AI publisher relationships
When I started researching these publisher relationships, one problem became obvious immediately.
A simple list of publishing companies isn't particularly useful for Digital PR.
“Future has an agreement with OpenAI” is interesting information.
But you don't pitch Future.
You pitch TechRadar or PC Gamer, or Marie Claire, or Tom's Guide.
The same problem appears with Hearst, Condé Nast, News Corp, Gannett, PRISA, and almost every large media organisation.
That is why the Spreadsheet database accompanying this article is built primarily around the publication/domain level, rather than only the parent company.
The first version contains 202 mapped publications or publishing assets.
Some appear explicitly in primary partnership announcements.
Others have been mapped because an agreement clearly covers the broader portfolio of the publisher that owns them.
And those two things must not be confused.
Therefore, I created an evidence classification.
An A-level record means that the individual publication or brand is explicitly named in primary documentation about the relationship.
A B-level record means that the publisher or network has an explicitly broad agreement and the publication has been mapped from that portfolio.
Imagine Future again.
OpenAI says its agreement involves more than 200 Future brands and specifically names titles such as Marie Claire, PC Gamer, TechRadar, Tom's Guide, The Week, Who What Wear and Cycling Weekly.
Those named publications are straightforward.
If another current Future property is included because the announced deal says the relationship covers Future's portfolio, that is a reasonable mapping, but it is analytically different from claiming that OpenAI individually named that title.
This distinction may seem pedantic. It isn't.
Media ownership changes. Brands are sold. Portfolios are reorganised. Agreements evolve. Precise contractual scopes are not always public.
A useful research database should preserve what we know, distinguish what we infer, and avoid quietly upgrading inference into fact.
Inside the Excel database
The workbook is divided into several layers because different questions require different levels of analysis.
The Dashboard provides the high-level view. It shows how many publications have currently been mapped to OpenAI, Google/Gemini, and Perplexity relationships, identifies multi-operator overlaps, and separates explicitly named publications from portfolio-derived records.
The Publications sheet is where the database becomes operational. Each row represents a publication or publishing asset and includes its domain, asset type, publisher, current ownership, market, language, and primary vertical. Separate columns identify relationships with OpenAI, Google/Gemini, Perplexity, and Anthropic, together with the type of relationship, evidence level, and primary verification sources. The same sheet contains practical fields for PR priority, outreach status, target journalist or desk, and campaign notes.
The Agreements sheet reverses the perspective. Instead of beginning with the publication, it begins with the relationship itself: AI operator, publishing organisation, announcement date, type of agreement, relevant AI surface, publications explicitly named, and source documentation.
That separation matters because the word “partnership” is dangerously vague. A direct real-time feed into Gemini is not the same thing as an AI-powered Google News experiment, and neither is automatically equivalent to a Perplexity publisher program.
The Expansion Queue identifies large portfolios where further mapping is useful, such as Future, Hearst, People Inc., or Gannett's USA TODAY Network.
Finally, the Methodology sheet explains how the records should be interpreted.
That may be less glamorous than the Dashboard, but it is one of the most important parts of the file.
A database without methodology gradually becomes folklore in spreadsheet form.
How I would use the database in a real campaign
I would not start by opening the spreadsheet.
Strategy begins with the brand.
What do we want to be known for? Which topics are genuinely connected to our expertise? What information can we contribute that does not already exist everywhere else? Which people inside the organisation can credibly represent that expertise? Which questions are journalists already asking?
In other words, we begin with many of the same ontology and information-gain questions that should already guide the website.
Then we map the relevant information ecosystem.
We identify the publications, journalists, trade media, newsletters, research organisations, podcasts and specialists that influence the conversation.
Only after creating that first universe would I intersect it with the AI publisher database.
Perhaps we discover that five of our thirty strong editorial targets have direct OpenAI relationships. Three participate in Google's publisher programs. Four appear in Perplexity's ecosystem. Two have relationships with more than one operator.
That information can change priorities.
But I would never remove a perfect editorial target simply because the corresponding AI column says “No”, nor would I pitch an irrelevant publication because three columns say “Yes”.
The logic should already be clear by this point: AI Ecosystem Proximity modifies a good target; it does not transform a bad target into a good one.
Multi-operator publications as research signals
The database also reveals another interesting category: publications with relationships across more than one AI operator.
The Washington Post, for example, appears in several of these ecosystems.
This does not imply some kind of cross-platform visibility multiplier when the publication mentions a brand.
What it tells us is simpler: several AI companies have independently considered a direct relationship with the same publication valuable enough to establish one.
That makes the publication interesting from an information-ecosystem perspective.
If it is also relevant to our topic, audience and campaign, it deserves closer attention.
This is where the spreadsheet works best: not as an answer machine, but as an investigation accelerator.
The website and Digital PR should share the same ontology
There is another operational consequence that is easy to overlook.
The topics we amplify through Digital PR should not exist independently from the authority architecture we build on our website.
Imagine that the website has been carefully structured around five major areas of expertise, while the PR team spends the next year promoting stories about fashionable subjects that have little connection with those areas.
SEO says the brand is about Topic A.
PR says it is about Topic Z.
Social changes the subject every week.
That is not distributed authority, but distributed confusion.
The ontology behind the website should therefore influence PR planning.
If we want to establish the company as an authority around sustainable aviation fuel, then our research, executive commentary, media pitching, partnerships, conference appearances and expert contributions should create legitimate associations around that semantic territory.
This does not mean endlessly repeating the same message but building coherence.
The website contains the structured architecture of that territory. PR extends it into independent information environments. Social carries it through people and communities. Brand marketing increases familiarity. Search captures and stimulates demand. AI Search becomes another interface through which the accumulated information may be retrieved and synthesised.
Different disciplines and tactics; one information ecosystem.
How to measure brand authority and amplification beyond backlinks
Once we accept this wider model, our measurement also needs to evolve.
Backlinks, referral traffic and search visibility remain useful, but they describe only part of the outcome.
I also want to understand whether relevant unlinked mentions are increasing; whether executives are repeatedly quoted around the subjects in which we want them to be recognised; whether original research is being attributed; whether first-generation coverage produces secondary coverage; whether the same brand-topic and person-topic relationships keep appearing in independent documents; and whether this broader presence coincides with stronger branded demand and greater visibility inside AI Search systems.
The interesting question is not merely whether the brand appears more frequently.
It is whether the brand increasingly appears in the semantic neighbourhoods it wants to occupy.
And when an AI system mentions the brand, we should investigate why.
Which sources appear to support the answer? Is the company's own website doing most of the work? Are third parties reinforcing the same associations? Is one piece of original research propagating particularly well? Is the brand simply one commodity recommendation among ten interchangeable alternatives, or is it connected with something distinctive?
That brings us back to information gain.
Visibility is useful. Distinctive association is stronger.
The goal is not to manipulate the model
We shouldn't approach Digital PR asking:
How can I manufacture enough mentions to make ChatGPT believe my brand is authoritative?
Apart from being strategically shallow, that question encourages precisely the kind of cargo-cult optimisation that SEO has suffered from for decades.
A better question is:
How can my company become a legitimate source of expertise and information that independent people have genuine reasons to reference?
Everything becomes more coherent once we ask the question that way.
Original research, expert commentary, long-term journalist relationships, institutional collaboration, and investment in people who can explain complex subjects clearly are no longer disconnected “brand tactics”. They are different ways of creating useful evidence around the same entity.
Technical SEO still matters because the brand needs a coherent, accessible, and semantically explicit home for its information.
Digital PR matters because the wider web needs independent reasons to recognise and discuss that information.
The two systems are complementary.
A famous company with an incoherent website makes machines—and people—work unnecessarily hard to understand what it knows.
A perfectly engineered website belonging to a company that exists nowhere else on the web remains largely self-referential.
The strongest position emerges when controlled and independent evidence points in the same direction.
Engineer brand authority at home. Earn it everywhere else.
Search itself is changing in ways that make this broader view increasingly relevant.
Google now explicitly documents that AI Overviews and AI Mode can use query fan-out, issuing multiple related searches across subtopics and data sources to build responses. Its newer guidance describes generative Search as grounded in pages retrieved from the core Search index and emphasises the value of unique, non-commodity content rather than content that merely repeats what already exists. Google's documentation for AI Search and its more recent guide to optimisation for generative Search are unusually explicit about these mechanisms.
ChatGPT Search similarly combines web search with relationships established directly with news and data providers, while presenting links to original sources. OpenAI's overview of ChatGPT Search describes this hybrid approach.
Publisher programs such as Perplexity's Comet Plus show that AI operators increasingly treat professional publishing as part of their information ecosystem rather than as something external to it.
In that environment, thinking exclusively about “my pages” becomes increasingly insufficient.
A technically strong website remains essential because it is where we structure our knowledge, clarify our entities and publish original information. But the wider information ecosystem determines whether that knowledge remains self-contained or becomes independently recognised, contextualised and propagated.
That is why I don't see Digital PR as something sitting outside SEO.
Nor do I see it as an AI Search trick.
I see it as part of the architecture of authority itself.
On our websites, we engineer what we know.
Outside them, we create the conditions for others to discover, test, contextualise, cite and amplify that knowledge.
And somewhere between those two environments, something much more valuable than another backlink can emerge:
a brand that the information ecosystem genuinely associates with expertise.
AI Search doesn't create the need for that.
It simply makes the need much harder to ignore.
Article by
Gianluca Fiorelli
With almost 20 years of experience in web marketing, Gianluca Fiorelli is a Strategic and International SEO Consultant who helps businesses improve their visibility and performance on organic search. Gianluca collaborated with clients from various industries and regions, such as Glassdoor, Idealista, Rastreator.com, Outsystems, Chess.com, SIXT Ride, Vegetables by Bayer, Visit California, Gamepix, James Edition and many others.
A very active member of the SEO community, Gianluca daily shares his insights and best practices on SEO, content, Search marketing strategy and the evolution of Search on social media channels such as X, Bluesky and LinkedIn and through the blog on his website: IloveSEO.net.
