What Tells a Search Engine What a Page Is About?

Mostly your words. The page title, the main heading, the subheadings and a clear answer near the top tell search engines and AI what a page is about. Structured data adds labels on top. It confirms who wrote the page, what kind of page it is and when it last changed.

Most advice puts it the other way around. It treats schema as the main signal and your writing as raw material. The words come first, so spend most of your time there. Structured data repeats what the words already say, in a form a machine reads. It can help a page get understood. It cannot make a dull page worth quoting.

The table below shows what a search engine reads on a page.

The short version

The title, H1, subheadings and early answer are words your reader sees. Structured data and links between pages are technical settings. The words matter more, yet most people spend their time on structured data.

What do search engines read first on a page?

They read the words people can see, starting with the title and the headings.

Microsoft says so in its guidance for AI search: “Your page title, description, and H1 tag (the top-level HTML heading) are important signals AI systems use to interpret purpose and scope.” All three of those are things a writer sets directly.

This is also why repeated titles cause trouble. If every page you publish has the same title, your strongest signal says nothing that tells them apart. Google asks for the opposite: “It's important to have distinct text that describes the content of the page in the title element for each page on your site”.

For creators, this mistake usually shows up in two predictable places.

The first is podcast or video episode pages. Many episodes are titled with the show name and a number, such as “The Creative Hour, Episode 47.” That tells a search engine which series the episode belongs to. It says almost nothing about what a listener will learn. A person searching for “how to price freelance branding projects” is not typing “episode 47.” The machine has no signal to match the question to your answer. The title helps file the episode. It does not help anyone find it.

The second is FAQ or advice pages where the creator's name becomes the default label. For example, a creator might put “Ask Alex” on every answer page, or repeat “Tom's Take” across dozens of posts. That can build your brand with a reader who is already on your site. To a search engine, it looks like fifty pages with the same title. Your strongest signal turns into noise, because nothing tells one page from another.

The fix is dull but it works. Put the question in the title. Word the title the way somebody would ask the question.

What is structured data, in plain English?

It is a small block of code in your page that labels what the page already says.

The labels come from a shared vocabulary called schema.org. The big search engines agreed on it so that everyone uses the same words for the same things. A page can say that it is an article. It can say who the author is. It can say when it was last changed. The usual format is called JSON-LD. It sits on the page as a block of code that visitors never see.

Google describes the purpose in one line: it “uses structured data that it finds on the web to understand the content of the page, as well as to gather information about the web and the world in general”.

Notice Google's word: understand. Your words still have to make the point. The label says what kind of thing it is.

One rule matters more than any choice of tag. The label has to match what a visitor can actually read on the page. Say you name an author who appears nowhere on the page. Or you give a date that differs from the one on screen. Then your labels describe a page that does not exist. Google ignores markup like that. If you use structured data, make sure it states the same things your text says.

Does structured data help you show up in AI answers?

It helps engines read your page correctly. On the evidence available, it does not earn citations by itself.

The engines have published what they do, and two teams have tested it.

Two engines say labels help machines understand a page. Two independent tests looked for more citations and found none. Both can be true. Being found and being picked to answer a question are two different things. Labels help an engine read a page correctly. They do not make it pick that page.

So add the labels because they describe your page accurately. They are not a way to buy visibility. Anyone selling schema as a route to citations is promising something no engine claims and no test found.

Labels also go out of date. FAQ markup used to be standard advice. In May of 2026, Google retired the FAQ rich results and has since removed its documentation altogether. If you added schema markup for that feature, it now does nothing in Google Search.

Which labels does a creator actually need?

Four, and most creators will only ever use three of them. Most creators use Person, ProfilePage and BlogPosting. QAPage is only for pages where readers post answers.

Google says ProfilePage is “designed for any site where creators (either people or organizations) share first-hand perspectives.” That describes what most creators run. A Q&A page is one “that contains data in a question and answer format, which is one question followed by its answers”. So it fits an answer page, not a general post.

Person is the label that does the most work. It lists the other places you exist: your channel, your profiles, your own site. That is how a machine connects three accounts to one person instead of reading them as three unrelated names.

Skip the rest. Product, review, recipe, local business and event markup all exist. None of them describes what you publish unless you really do sell products, write formal reviews, or run a venue. Marking a blog post as a product does not upgrade it. It describes a page that is not there.

What does this look like on a real creator page?

Here is a real one, read live on October 30, 2026.

The page is one answer in Dr Mitchell Levy's library on iSonic. A visitor sees the question as the H1: “What does Dr Mitchell Levy mean by 'living your values' in leadership?” His answer sits below it. The markup says the same thing in labels:

{
  "@type": "QAPage",
  "mainEntity": {
    "@type": "Question",
    "name": "What does Dr Mitchell Levy mean by 'living your values' in leadership?",
    "acceptedAnswer": {
      "@type": "Answer",
      "text": "Hi, I'm Dr Mitchell Levy from Credibility Nation. How can I help? When talking about 'living your values' in leadership, the focus is on who you are, not just what you do or what you achieve. ...",
      "author": {
        "@type": "Person",
        "@id": "https://isonic.ai/creators/mitchelllevy/#person",
        "name": "Dr Mitchell Levy"
      },
      "datePublished": "2026-10-01T02:03:41Z",
      "dateModified": "2026-10-06T02:06:02Z",
      "citation": [
        "https://www.youtube.com/watch?v=HhogdAS4-3I",
        "https://www.youtube.com/watch?v=ce5zE34BiAs",
        "https://www.youtube.com/watch?v=eXcNCNgdmsE"
      ]
    }
  }
}

In plain English, that says five things. This page is one question and one answer. A named person wrote it. Here is when it was published. Here is when it last changed. Here are the three videos the answer came from.

That last part is the one creators usually have no way to state. The answer names the videos it came from. That spells out the link between the written answer and the original work.

The other half is on his profile page, one of many in the creator library. That page carries ProfilePage markup. It also carries a Person whose sameAs list points at his YouTube channel and his LinkedIn profile. That Person has an identifier, and the same identifier appears as the author on every answer page in his library.

That shared identifier turns a folder of separate pages into one person with a body of work. Without it, you have a hundred pages that each happen to mention the same name.

How do you add structured data to your own site?

Start by checking what you already have, because most platforms add some of this for you.

1. Run one of your existing pages through Google's Rich Results Test or the Schema.org validator. If your platform already adds an Article label with an author, you have less to do than you think.
2. Add a Person label to your about page, with your name and a sameAs list of every profile you run. This is the label that ties everything else together.
3. Make sure each post carries an author and both dates. Published and updated are separate facts.
4. Test again after each change, on a real page rather than a sample.
5. Check the labels still match the visible text. People often skip this step after a redesign. Bylines move or disappear, but the markup still lists them.

On WordPress, Squarespace, Wix and similar platforms, adding the author and dates is usually a setting. You do not need code. A plugin usually covers the Person label. You do not need a developer. You do need to read the validator output instead of trusting what the plugin says about itself.

Do two pages on the same topic confuse AI?

Not if each one answers a different question. Two pages on the same subject are fine. Two copies of the same answer are not.

Google is relaxed about one page covering several things. Its guidance says “Google systems are able to understand the nuance of multiple topics on a page”. What it warns against is the same content at more than one address. Google says “having duplicate content can be a bad user experience and search engines might waste crawling resources on URLs that you don't even care about.”

So two posts about espresso machines are fine. Two posts that both answer “is the Gaggia Classic worth it” are really one post and a copy. You are asking an engine to pick between them for you.

Give each page its own question and its own title. If you cannot think of a separate question for the second page, do not write it.

What can nobody promise?

A citation, a ranking or a mention. Markup cannot promise them, and neither can any tool, iSonic included.

Google will not even promise the labels get used. Its structured data policies say it “does not guarantee that your structured data will show up in search results”. Google says this about its own features, even for pages with correct markup.

The two tests above looked for more citations and found no gain beyond normal variation. Remember that when someone quotes you a percentage. The labels make your page easy to read and your authorship clear. That is all they do.

Where does iSonic fit?

iSonic writes this markup for you, from content you already published.

iSonic is a GEO platform that turns a creator's existing content into structured, AI-readable pages so AI assistants can find it and credit it to the creator. In practice that means an AI Bio Page carrying ProfilePage and Person. It also means GEO Questions carrying QAPage, with you as the author and your source videos cited. That is the same markup shown above.

Get your free AI Bio Page

Quick answers

Is structured data required for AI search?

No. Google says structured data is not required for generative AI search, and that there is no special schema.org markup you need to add. It also says the labels are still worth keeping as part of normal SEO. Use them to label your page accurately. Leaving them out is not a failure.

Do I need a developer to add structured data?

Usually not. A plugin or a settings panel covers it. The real work is checking the result in a validator. Make sure the labels match what a visitor sees on the page.

Does Wix, Squarespace or WordPress add schema for me?

Partly, and how much varies by platform and theme. You may need to add a Person label to your about page yourself, with links to your other profiles. Test a live page rather than assuming.

Can you give me an example of schema markup?

The shortest useful one is a Person label on your about page. It holds the type, your name, your site, and a sameAs list of your other profiles. The example above shows a fuller version, with a question, an answer, an author, and the videos the answer came from.

Should the answer come before the explanation?

Yes. Put the direct answer in the first two or three sentences, then explain it. An assistant looking for a passage to quote takes the part that answers the question. A page that saves its conclusion for the end has nothing to offer near the top.

Your words tell engines what the page is about. The labels just make sure the right name is attached when somebody quotes them.

Get your free AI Bio Page