AI engines do not read your website the way a visitor does. They read how it is put together. If your website structure for AI search is a five-page brochure with “learn more” buttons, there is nothing for ChatGPT or a Google AI Overview to understand, so there is nothing for it to cite. Here is the process I teach and use: a mind map, then hubs and silos, then the signals that tie them together.
KEY TAKEAWAYS
- Start with a mind map of your business, compared against the biggest site in your space, not another small site.
- Group pages into hubs: parent, children and grandchildren. The children point up and push the parent up.
- Every link needs a keyword anchor. “Learn more” tells an AI engine nothing.
- Put an FAQ on every page and use the schema for your business type. Then test it.
- Publish a hub all at once, not one page a month.
Why structure decides whether AI engines find you
I have been teaching SEO since 2003, and the more things change, the more they stay the same. Links matter. Authority matters. But in the AI era, technical SEO is what wins, and the biggest piece of technical SEO is structure: how your pages are organized, what sits under what, and how the pages link to each other. Traditional search and AI search read the same site. Google Search, Google AI Overviews and ChatGPT all have to work out what your site is about before an AI system can cite it, and structure is how you give them that clarity.
Here is the part some people do not like hearing. Structure lives on your website, so your platform matters. Wix, Squarespace, WordPress.com and the website-in-a-box builders do not give you the control you need over URLs, schema and page hierarchy. I build on self-hosted WordPress.
“If you have a five-page website, you might as well just hang it up.”
Beth Guide, Website Structure for AI Search (Internet Marketing Clinic, October 2026)
Someone told me 100 pages sounded like overkill. AI does not think so. A well organized site of a hundred pages, all on topic, gives an answer engine a full picture of what you do and where you do it.
Step 1: map your site against the biggest site in your space
The first thing I do on any build is a mind map. I used to draw it in pencil on a very large sheet of paper. Now I build it in Claude, because I can feed it real websites and get back a working plan I can add notes to and mark complete as pages go live.
Vertical Web is really three businesses in one: SEO, web design and hosting. So I built three mind maps. An accounting firm is the same story with accounting, tax preparation and bookkeeping. Start by writing down your highest level services. Each one becomes a hub.
Then pick sites to compare structure against. Not another five-page site. Pick the biggest company you can find in your space. I picked two agencies whose structure I liked, one in Pennsylvania and one in Conroe, and asked Claude three questions: what pages do they have, what pages do we have, and what pages are we missing?
The answers were humbling. My industries list started at 12. The AI, comparing it against the sites we have actually built, came back with 27. I also had more than 60 location pages, and some were so alike they overlapped by close to 40 percent. Pages in a set like that should share only a few percent.
WATCH THE CLASS
Website Structure for AI Search (Internet Marketing Clinic, October 2026). Beth Guide teaches this live, most months, with HCC and SCORE.
Step 2: build hubs and silos so every page supports its parent
A hub is a parent page with its children and grandchildren underneath it. On my site, Areas We Serve is the parent. The areas of town are the children. Under them sit grandchildren such as the Galleria, Midtown, the Heights and the Medical Center, all linked to each other and back up.
“all the children point to the parent and push the parent up”
Beth Guide, Website Structure for AI Search (Internet Marketing Clinic, October 2026)
Think of the houses in Galveston up on cinder blocks. The higher you want the house, the more blocks you put under it. Supporting pages are the blocks. A hub built this way is a self-contained mini website that would make sense on its own domain.
That real estate site is the cleanest example I have. It went up on July 1. Every county, suburb and subdivision has its own page pointing back to its parent, plus guides for buyers, sellers, relocation and school districts. Within its first few months it was ranking for more than 700 keywords and had 27 AI listings. When the agent asked for a section on a fast-growing area north of Houston, I published the county pages, the town pages and the guides at the same time. That section was ranking within about a week, ahead of news sites.
To see this on your own site, the SEO tools in elev8tor include an interactive site map that draws your topic hubs, marks in red where pages do not link up, and lists the gap pages to build. You can print the outline and hand it to whoever writes the pages.
Step 3: the signals that tie a hub together
- Anchor text. A button that says “learn more” does not tell anyone what the next page covers. The words “luxury homes” linked to the luxury homes page do. I go deeper on this in internal linking for AI search.
- Header tags. A heading that says “Look at the color” could be about rainbows, a steak or a car. Put the keyword in the heading.
- Title tags. Still the first thing I fix.
- Breadcrumbs. AI drafting tools often forget them. They help engines follow the path from grandchild to parent.
- Related pages. A block at the bottom that links siblings and the parent.
An SEO audit catches the mechanical side of this: missing titles, weak H tags, pages that cannot be indexed. Each issue comes with How to fix in plain words, so you are not decoding a report.
Step 4: FAQs and structured data on every page
Google keeps saying FAQs and schema matter less for its snippets. Maybe. What I see is that AI Overviews and the other AI engines draw on FAQs when they assemble answers and pick a citation, so I put an FAQ on every page. Do not drop something just because Google says to.
“Schema is kind of like a spare tire.”
Beth Guide, Website Structure for AI Search (Internet Marketing Clinic, October 2026)
You do not know you need it until it is not there. Schema is structured data: code that tells machines what kind of business you are: real estate agent, dentist, lawyer, travel agency. A travel agent I work with had plenty of travel guides and no travel agency schema, so Google could not tell whether he was an information site or an agency. Travel agency schema on every page settled it. On WordPress, Rank Math handles it. Then test it, because generated schema can carry an old address or skip the country code.
elev8tor checks findability too: noindex tags, canonicals pointing elsewhere, and whether robots.txt lets Googlebot, OAI-SearchBot, GPTBot, PerplexityBot and ClaudeBot in. That is part of our AI search optimization work.
How to check your structure this week
- List your top level services. Each one is a hub.
- Pick the biggest site in your space and compare page lists. Write down every page you are missing.
- Sort the missing pages under a parent. If a page has no parent, ask whether it belongs on the site.
- Replace every “learn more” with a keyword anchor.
- Add an FAQ and the right schema to each page, then test the schema.
- Run each new page through the content scanner before it goes live, so it is measured against today’s top 10 instead of a guess.
Content comes after structure. If the structure is wrong, the best copy in the world will not do you a bit of good. Once it is right, the next job is what goes on each page: entity coverage and originality, which I cover in separate posts. If you manage many client sites, the same process runs across all of them in our agency SEO platform. And if you want the background on how this tool came out of our own client work, read why elev8tor.
FAQ
Frequently asked questions
There is no fixed number, but a five-page site gives Google Search and AI Overviews very little to work with. Build every page your hubs need, compared against the biggest site in your space. For a competitive local service business that often means 100 pages or more, and even a hyper niche manufacturer can need several hundred.
A silo is a group of related pages: a parent page, the child pages under it, and grandchild pages under those. The pages link to each other and back up to the parent with keyword anchors, so the parent is supported from below.
Yes. AI engines draw on FAQ content when they build answers, and schema tells engines what type of business you are. Add both to every page where they fit, then test the schema so it carries the right name, address and business type.
Yes, with supervision. Give it your top level services and two or three larger sites in your space, ask what pages they have that you do not, then correct the result. Beth once got an automotive section written for car dealers when she wanted auto repair shops.
Check your own page against today's top 10
The free demo gives you 5 real Content Scanner scans on your own pages. No card, no sales call.
Related posts and pages
- Internal linking for AI search: the exact sentence and anchor for every link
- Keyword density vs entity coverage: why fixed keyword counts are the wrong target
- AI search optimization: how elev8tor checks AI findability and citations
- SEO tools: site map, audits, rank tracking and research in one place
elev8tor.ai is developed by Vertical Web and Beth Guide, drawing on more than 20 years of SEO work.
- Internal Linking for AI Search: The Exact Sentence and Anchor Text - October 23, 2026
- How to Prove SEO ROI with GA4 Lead Tracking - October 22, 2026
- Keyword Density Is the Wrong Target: Use Entity Coverage Instead - October 21, 2026