Okay, I need to talk about something that's been nagging at me since I fell down a research rabbit hole at an hour I'm not prepared to disclose. It's about how AI search engines decide who to cite — and it has surprisingly little to do with your website and a lot more to do with you. Your actual name. Attached to your actual face. Saying things you actually know.

I realise that sounds like a motivational poster in a WeWork. Bear with me.

For twenty years, the unit of SEO was the page. You optimised a URL, earned links to it, and waited for it to climb the rankings like a very patient mountaineer made of metadata. The person who wrote it was irrelevant — the algorithm didn't care if it was penned by a Nobel laureate or a particularly ambitious golden retriever with a WordPress account.

Answer engines have changed the question. When ChatGPT, Perplexity, or Google's AI Overviews assemble a response, they're selecting roughly five sources, synthesising them, and presenting the result in one confident voice. That compression forces an editorial decision ranking never did: before a model restates a claim as its own, it has to decide whether the source is credible enough to stand behind.

And the most reliable way any system has ever made that judgement is by asking who is making the claim, and whether anyone independent vouches for them.

Which, if you've spent the last decade optimising pages rather than people, is the kind of revelation that makes you want to lie down.

Here's the thing — these models weren't programmed to prefer named authors. They were trained on the internet, and the internet already encoded two decades of human judgement about which sources to trust. Claims that traced back to identifiable experts carried more weight in the training data than claims that materialised from the content void like a press release from nobody.

The model absorbed the pattern along with the prose. It's reflecting the same trust signals search engines spent twenty years refining under E-E-A-T — experience, expertise, authoritativeness, and trustworthiness. (A framework that sounds like it was named by a committee that ran out of vowels.)

The evidence has become hard to wave away. Studies of citation patterns across ChatGPT and Perplexity found that pages without a named author are roughly forty per cent less likely to be cited than equivalent content with a verifiable byline. Google added a dedicated authors section to Search Central at the start of 2026. One analysis of AI Overview citations found the overwhelming majority came from sources exhibiting strong E-E-A-T signals.

Having a credible, identifiable human behind a claim has moved from "nice to have" to something closer to a prerequisite for being cited. And "Brand Name Content Team" doesn't count, in case you were wondering.

Experience — genuine first-hand knowledge of a product, a category, a problem — is precisely the input a language model cannot synthesise from its training data, because by definition it doesn't exist there yet. This is the information gain that keeps coming up: the unique data, the first-party observation, the expert judgement that adds something the model couldn't generate on its own.

As competent-but-undifferentiated content floods the web (thanks, generative AI — eating your own tail and somehow still hungry), first-hand expertise becomes the scarce and defensible input. And first-hand expertise has an owner. It belongs to a person.

Not a brand guidelines document. Not a content calendar. A person.

We've talked before about how the further a voice sits from direct brand control, the more answer engines tend to trust it. A brand asserting it's the best in its category is the least persuasive signal available — it's the corporate equivalent of marking your own homework and giving yourself an A+. This is why community platforms and third-party voices punch above their weight in AI citations.

But here's where it gets interesting. The founder — or the in-house expert more broadly — occupies a weirdly productive position on that trust spectrum. They're internal enough to possess genuine first-hand knowledge: the operator who's run the supply chain, the consultant who's fielded the same objection four hundred times, the product lead who knows why version 3.2 broke. That's experience no synthesised content can replicate.

But they're also external enough to be corroborated. A founder who writes under their own name, appears on podcasts, speaks at events, contributes to trade publications, and maintains a substantive professional profile becomes verifiable across dozens of independent surfaces. The claim is no longer "brand says brand is good." It's "a named expert with a traceable record says this, and the wider web confirms the expert is real."

That combination — first-hand authority plus independent corroboration — is exactly what an answer engine is built to reward. The founder is the highest-trust human asset a brand actually controls, and the one whose expertise is least reducible to template.

(If reading that made you slightly anxious about getting on camera, good. That discomfort is load-bearing.)

Here's where most people go wrong: they think this is about slapping a name at the bottom of a blog post and calling it founder-led content. In an entity-first search environment, the person isn't a byline — they're an entity. A node in the same semantic network that holds the brand, its services, and its category, with its own attributes, associations, and credibility score.

Building that entity looks less like content production and more like disambiguation. Person schema linking each piece of content to the author's full entity. SameAs connections tying that entity to a LinkedIn profile, a Crunchbase listing, a conference speaker page, wherever else they exist. Each consistent connection lowers the cost, in the model's terms, of confirming that this named expert is a single, real, coherent person rather than an ambiguous string of characters that could refer to anyone. (There are a lot of John Smiths. The models know this.)

But the more durable half of the work happens off your own domain. During training, models learn associations between author names and topics from millions of pages. An expert who talks consistently about one narrow area, across enough independent surfaces, becomes the entity the model reaches for when that subject comes up. Some practitioners call this multi-touch authority — the idea that an author's standing is assembled from their footprint across podcasts, guest articles, cited case studies, and social presence, rather than from on-page keywords.

LinkedIn has become one of the most-cited domains for professional and B2B queries across major AI platforms — making a substantive founder presence there closer to infrastructure than vanity. YouTube transcripts feed generative summaries. Community platforms reward genuine participation. The same expertise has to show up native to each surface, because each engine assembles its answer from a different pool of sources.

The reason to treat founder-led expertise as a foundation rather than a campaign is that it compounds in ways few other AEO activities do.

As generative tools push the cost of competent content towards zero, every category converges on the same fluent, forgettable middle. What survives is the thing models can't generate: a specific human point of view grounded in real experience. Meanwhile, audiences grow warier of content they suspect was machine-produced, which makes signals that read as unmistakably human — a named person, a track record, a willingness to be wrong in public — scarcer and more valuable.

Both dynamics are reinforced by how entity recognition works. Models reinforce what they already know. An expert the system has learned to associate with a topic gets surfaced again, which generates more mentions, which deepens the association. Early, consistent investment widens the advantage over time, and latecomers face a steeper climb.

Which is a polite way of saying: the best time to start was six months ago, and the second best time is before you finish reading this paragraph.

Here's where theory meets "I have a business to run and seventeen things on my to-do list already." The good news is that this doesn't require a media empire. It requires consistency and a willingness to be a real person on the internet, which is admittedly harder than it sounds.

Pick your person (or people). Start with the founder or the most credible subject-matter expert you have. Not the most polished — the most knowledgeable. The person who can explain why your approach works differently from everyone else's, and who has the battle scars to prove it. If you're a SaaS founder who's spent three years solving a specific integration problem, that's your angle. If you run a law firm that specialises in a niche area, the senior partner who's tried forty cases in that space is your asset.

Build the entity, not just the content. Add Person schema to every piece of content that person authors. Connect their author profile to their LinkedIn, their conference bio, their guest articles — anywhere they exist independently. The goal is to make it trivially easy for a model to confirm this is one real, coherent expert, not just a name on a page.

Get off your own domain. This is the part that feels uncomfortable, but it's the part that actually builds authority. Appear on podcasts in your space. Write for trade publications. Post substantive takes on LinkedIn — not "thought leadership" platitudes, but the specific, opinionated, experience-backed observations that only someone doing the work could make. Each independent surface that corroborates your expertise is another data point the models learn from.

Distribute the expertise. Don't bet everything on one person. Your head of product, your operations lead, your senior consultant — they each hold first-hand knowledge no synthesised content can replicate. Put their names on it. Give them bylines. Spread your authority across several entities so you're not one departure away from an entity-graph-shaped hole in your visibility.

Audit what you already have. Most businesses are sitting on content that's published under "Admin" or "The Team" or nobody at all. Reattributing existing high-performing content to named authors with proper schema is one of the fastest wins available. It won't make the news, but it makes the models' job easier, and making the models' job easier is basically the whole game now.

For twenty years, the brand was the unit of authority in search — the thing that accrued links, ranked, and got remembered. Answer engines are quietly promoting a smaller unit beneath it: the identifiable, corroborated person.

If the models are learning to trust people more readily than pages, the organisations that thrive under AI search might be the ones that stop asking "how do we make our brand more authoritative?" and start asking "whose expertise can we afford to make visible?"

The answer, inconveniently, is probably "yours." And yes, that means getting on the podcast.

Jo Lambadjieva is an entrepreneur and AI expert in the e-commerce industry. She is the founder and CEO of Amazing Wave, an agency specializing in AI-driven solutions for e-commerce businesses. With over 13 years of experience in digital marketing, agency work, and e-commerce, Joanna has established herself as a thought leader in integrating AI technologies for business growth.

😍It was great, all I needed

🙂Good, not great

🙄It sucked