How AI Overviews Decide Who Counts as an Expert
How AI Overviews Decide Who Counts as an Expert in Your Field
Last updated: July 2026
For almost every professional query someone types into Google in 2026, an AI Overview appears at the top of the page and hands the searcher a summarized answer, plus a short list of citations. Those citations are the new front page of the internet. If a doctor, attorney, financial advisor, or executive shows up inside that citation stack, the searcher reads them as the expert on the topic. If they don’t, the searcher never scrolls far enough to find them.
The question we get from clients almost every day is a version of the same one. How does Google decide who counts as an expert? Why is that competitor being cited and I’m not? And what would it take to change that?
This post is the honest answer, based on Google’s own guidelines, current citation research, and the pattern we see across every AI reputation engagement.
Who is actually getting cited right now
Before we talk about signals, it helps to look at what is actually inside AI Overview citation stacks in 2026.
The largest cross-platform citation study to date, run by 5W Public Relations and released in May 2026, synthesized more than 680 million citations across ChatGPT, Google AI Overviews, Perplexity, Gemini, and Claude. It found that the top 20 domains account for 66 percent of all AI citations across the major answer engines. Reddit sits at number one across every engine. Wikipedia, YouTube, LinkedIn, and Forbes round out the top five. A Search Engine Land analysis of the same pattern showed Reddit alone is cited roughly 40 percent of the time inside major LLMs.
That is the ecosystem your name has to earn a seat inside. When Google’s AI Overview picks an expert to cite for a query like “best way to protect a corporate CEO from an online smear campaign” or “how to remove a defamatory article about a doctor,” it is choosing from a small set of trusted publishers, plus the personal profile pages and named authors that those publishers link out to.
Being cited matters. An April 2026 Seer Interactive study found that pages cited inside an AI Overview earn roughly 120 percent more organic clicks per impression than uncited pages on the same query. And Ahrefs found that AI Overviews correlate with a 58 percent reduction in click through rate for uncited top ranking pages. Being cited is the new page one. Being uncited is worse than being on page two used to be.
Almost half of American adults now use AI chatbots, and 60 percent read AI generated search summaries, according to the June 2026 Pew Research Center “Americans and AI 2026” report. Twenty four percent of US adults use them daily. If you are a professional whose clients find you through search, that is where you have to be visible.
What Google actually means by “expert”
Google publishes a document called the Search Quality Rater Guidelines. It is the closest thing to a rulebook for what Google considers high quality content, and Google updated it twice in 2025 to make the framework tighter. The current version of the guidelines PDF uses a framework called E-E-A-T. It stands for Experience, Expertise, Authoritativeness, and Trust.
Each letter is a specific question the rater is asked to answer, and by extension what the AI Overview citation system is trained to detect.
Experience asks whether the author has firsthand involvement with the topic. For a professional, that means content written from actual client work, not from a keyword tool.
Expertise asks about credentials, education, and demonstrated competence. Board certifications, bar admissions, published work, and years of practice all count.
Authoritativeness asks whether other trusted sources recognize this person or brand as a leader in the field. Coverage in respected publications, citations in professional literature, and repeated inclusion in industry lists all move this signal.
Trust is the umbrella that ties everything together. It asks whether the source is transparent about who they are, whether the content is accurate, and whether the site is safe.
Every AI Overview citation the system produces is filtered through this framework. A 2026 analysis published by Wellows found that 96 percent of AI Overview citations come from sources with verified E-E-A-T signals, and that pages with named, credentialed authorship are 35 percent more likely to be featured than pages without.
The practical takeaway is that E-E-A-T is no longer a soft SEO best practice. It is an active gatekeeper. Content without a real, verifiable expert attached to it is filtered out before the ranking algorithm even considers other factors.
The four signals AI Overviews actually read
E-E-A-T is the framework. The signals below are how the framework gets applied at machine scale. These are the four things AI answer engines actually look at when they decide whether you count as an expert in your field.
1. Named authorship with a verifiable bio
The single biggest change in the last two years is that anonymous or ghost bylined content no longer competes for AI citations. When Google’s system evaluates a page, it looks for a named author, a real bio, disclosed credentials, and a profile photo that resolves to a real person. Google’s own helpful content documentation tells creators to make it clear “who” is behind the content and to publish author information that makes expertise obvious to a reader.
For a professional, that means every piece of content on your firm’s website needs a byline, a link to a real bio page, credentials listed, and a photo. Not a stock illustration. A real photo of the real person. The bio page should include your license or bar or board information, your firm affiliation, your published work, and a clean, current headshot.
2. Author entity resolution in the Knowledge Graph
Google’s Knowledge Graph is the internal database Google uses to understand real world entities. People, companies, places, and concepts each get their own entry. When Google’s AI Overview system encounters your name on a page, it tries to resolve that name against a Knowledge Graph entity. If it finds one, and the entity has a strong citation trail behind it, your citation weight goes up. If it does not find one, your name is just a text string on a page.
For most professionals, the fastest way to earn entity recognition is through Wikidata, the open knowledge base that feeds directly into Google’s Knowledge Graph and into large language model training data. A well formed Wikidata entry with references to press coverage, licensing databases, and other authoritative sources gives Google a stable identity to attach your citations to. It is the foundation layer.
Wikidata has notability rules that are less strict than Wikipedia’s, but it still requires external references. That means the sequence matters. You need press coverage and third party mentions first, and then the Wikidata entry becomes the anchor that pulls it all together.
3. Structured data with schema.org Person and sameAs
Schema markup is machine readable code embedded on a webpage that describes what the page is about. For expert citations, the two schemas that matter are Article schema and Person schema, connected by an authored by relationship.
The single highest leverage property inside Person schema is called sameAs. It is a list of URLs pointing to other authoritative profiles of the same person. Your LinkedIn. Your Wikidata entry. Your bar directory profile. Your Healthgrades page. Your Crunchbase profile. Each sameAs link tells Google’s system that all of these pages describe the same real person, and each link raises what researchers call your entity confidence score.
An analysis published on Search Engine Land noted that schema is one of the clearest ways for a publisher to disambiguate an entity, and cited studies have found that 81 percent of pages cited in AI search responses include schema markup. The schema.org Person specification is public and free to implement. Most reputable publishers already know how. Most professional websites still do not have it configured correctly.
4. Third party mentions on cited source domains
The final signal is the hardest one to move but the one that matters most in the long run. Because AI Overviews pull disproportionately from a small group of trusted domains, the fastest way to become a cited expert is to appear on those domains as a quoted source.
That means guest columns in industry publications, cited quotes in journalist reporting, appearances in Forbes contributor programs where the byline standards still allow it, speaking slots on YouTube channels that have real subscriber authority, and answers in Reddit threads inside communities relevant to your field. It also means being a source that reporters can find and quote when a news cycle breaks in your industry.
None of that is a hack. It is the same reputation building that professionals have done for decades, translated into a channel mix that Google’s AI Overview system now watches.
What this means if you are a professional in an active practice
For most attorneys, doctors, financial advisors, and executives we work with, the pattern is the same. Their competitor is being cited inside AI Overviews for high value queries in their city or specialty. They are not. The difference is almost never that the competitor is better at their job. The difference is that the competitor’s name resolves to a coherent entity across LinkedIn, their firm bio, Wikidata, a Google Business Profile, at least one press mention on a top 20 cited domain, and a website that carries proper Person schema pointing at all of it.
The competitor has been made legible to the system. You have not been.
The good news is that legibility is fixable. It takes work, but the work is knowable and repeatable. It is also the work our AI Search Reputation Management practice does every day.
The mistakes we see most often
A few recurring errors we see when clients try to solve this themselves.
Skipping the bio and photo. A stock photo and a two line “about the author” block does more harm than good. If Google’s system cannot verify who the author is, the page gets a lower expertise score than if there were no byline at all.
Publishing without a Person schema on any of it. Firms will invest six figures in a website redesign and still ship without Person schema. That is money left on the table. The technical implementation cost is trivial. The signal cost of leaving it off is enormous.
Chasing volume instead of authority. Publishing forty thin blog posts a month does not build expertise signals. Publishing eight strong pieces from a named expert, each linked from a real bio, does. Google formally added a “scaled content abuse” category to the rater guidelines in September 2025 for exactly this pattern.
Ignoring the Knowledge Graph layer. Executives will spend a year on media outreach and never once try to convert that coverage into a Wikidata entry or a properly linked LinkedIn profile. The entity layer is what carries the coverage forward into every future AI search.
Treating AI reputation and traditional SEO as different projects. They share the same underlying entity graph. Investing in one without the other leaves half the return on the table.
A framework you can actually use
Here is the sequence we run for individual and executive clients who want to become the cited expert in their field.
- Audit every AI Overview and AI chatbot answer for your top 20 client search queries. Note who is cited and who is not, and note which domains those citations come from.
- Fix the author layer on your own site first. Real bios, real photos, real credentials on every piece of content, plus a Person schema block with a sameAs list of every authoritative profile of you that already exists.
- Cross link the profile stack. LinkedIn should link to your firm bio. Your firm bio should link to your published work. Your Google Business Profile should link to your website and to your firm bio. Consistency across profiles is what teaches the entity graph.
- Build a press track. Even one placement per quarter in a top 20 cited domain will move your entity score meaningfully, provided the byline or the quote credits you by full name and title.
- Submit a Wikidata entry once you have the external references to support it. Use the press coverage, professional licensing directory URLs, and any speaking or publishing history as citations.
- Publish original expert content on your own domain that a journalist could source. Data you collected. Case outcomes you can share. Predictions you are willing to be measured on later. That is the raw material AI Overviews want to cite.
- Measure again. Rerun the same 20 queries every month for the first six months, then quarterly. Track which citations you have earned and which are still going to competitors.
We run this sequence for individual reputation clients and executive crisis clients on a subscription basis, and for company clients through the business reputation practice. If you would rather understand the process yourself first, that is what this post is for.
Frequently asked questions
How long does it take to start showing up in AI Overviews as an expert?
If you already have a base of press mentions and a real professional profile, the technical fixes can start moving citations within four to eight weeks. If you are starting from an empty slate, the honest answer is six to nine months of consistent work before AI Overviews begin citing you at scale.
Does paying for a Wikipedia article help?
We would not recommend it. Wikipedia has strict rules against paid editing, and undisclosed paid contributions get reversed and can damage your Knowledge Graph presence in the process. Wikidata is the right layer to invest in first. Wikipedia should come organically, if at all.
Do AI Overviews still cite small websites, or is it only major publishers?
They cite both. The top 20 domains carry the majority of citations, but individual expert quotes and named author bylines on smaller professional sites are cited constantly, especially for local and specialty queries. That is the door you walk through.
Can I get removed from an AI Overview that misidentifies me?
Sometimes yes. Google offers a “Results about you” flow and a legal removal path, and each major AI engine has its own reporting mechanism. We walk through the chatbot correction process in detail in our post on how to fix wrong information about you in ChatGPT, Gemini, and Perplexity, and the broader content workflow is covered in our content removal practice and our suppress negative search results service.
What if a competitor is being cited using content that is actually mine or that misrepresents me?
That is a hybrid problem. You need to earn citations for yourself and you need to correct or suppress the ones that misrepresent you. If the misrepresenting content is defamatory, that is a separate track we cover in how to remove defamatory content from Google. We run both tracks together when clients come in with that situation.
Where to go from here
The professionals who will be visible in AI search over the next five years are the ones who treat expertise the same way Google’s system treats it. As something a machine has to be able to verify, cross reference, and cite with confidence. The credential is not enough. The website is not enough. The one Forbes mention is not enough. The stack is what matters.
Every DCM engagement carries a guarantee on outcomes. If we take you on and the AI Overview citation stack does not move for your key queries within the engagement window, we do not walk away with your money and a shrug. That is a promise we can only make because we have run this playbook enough times to know exactly what it takes.
If you want to talk through where you stand today, book a free consultation or head over to the homepage to see the full scope of what we do. And if you want to keep reading, our earlier posts on fixing wrong information about you inside ChatGPT, Gemini, and Perplexity and how to remove defamatory content from Google cover the two adjacent problems most professionals hit alongside this one.
The engine is not going to slow down. The professionals who become legible to it first are the ones who get to keep their phones ringing.
