Bots read your site first: schema, entities and clean HTML
Before a customer finds you on Google or ChatGPT, a bot has already read your site. See what it “sees,” how to help it understand who you are and why this is now a web development matter.
Your first visitor isn’t human
According to Imperva’s 2026 report, more than 53% of internet traffic now comes from bots. Many of them are malicious, but among them are also the ones that decide whether a customer will ever find you: Googlebot, Bingbot and the crawlers of AI tools, such as ChatGPT’s OAI-SearchBot and PerplexityBot.
These bots don’t see your site’s colors, photos or animations. They read the code. If the code is sloppy, an impressive site can look almost empty to them. And anything they don’t understand, they won’t show on Google or in AI answers.
What does a bot “see” when it opens your site?
A bot reads the page’s text, headings, links and structured data. It struggles, or fails completely, with two things:
- Text inside images: A banner with your services written over a photo is, to a bot, just a photo.
- Content that loads with JavaScript: According to an analysis by Vercel, the crawlers of ChatGPT, Claude and Perplexity don’t execute JavaScript. If your site “builds” its content in the visitor’s browser, these tools don’t see it at all. Googlebot does read it, but it takes an extra step and more time.
A simple test: open a service page and press Ctrl+U (Cmd+Option+U on a Mac) to view the source code. If you can’t find the page’s text there, most AI bots probably can’t find it either.
Clean HTML: the foundation you can’t see
“Clean” HTML isn’t about how pretty the code looks. It’s how a machine understands your page’s structure at first glance:
- Proper heading hierarchy: One H1 per page, with H2s and H3s in a logical order. For a bot, the headings are the table of contents.
- Semantic elements: Tags like <nav>, <main>, <article> and <footer> show which part is the menu and which is the actual content.
- Real links: Links with a proper <a href>, not buttons that open pages only through JavaScript. Otherwise, bots can’t find the rest of your pages.
- Image descriptions (alt text): The only way for a bot to “see” what a photo shows.
Schema markup: tell Google who you are
Schema markup, also called structured data, is a piece of code, invisible to the visitor, that explicitly describes what’s on the page. It uses the Schema.org vocabulary, created jointly by Google, Microsoft, Yahoo and Yandex, and Google recommends writing it in JSON-LD format. Instead of letting the bot “guess,” you tell it: “This is a dental practice, at this address, with these opening hours.”
{
"@context": "https://schema.org",
"@type": "Dentist",
"name": "Example Dental Practice",
"address": "100 Kifisias Avenue, Marousi",
"telephone": "+30 210 0000000",
"openingHours": "Mo-Fr 09:00-20:00"
}
The most useful types for most businesses are Organization or LocalBusiness (with more specific types, such as Dentist or MedicalClinic), Product for e-shops, Article for articles and BreadcrumbList for navigation.
What it doesn’t do: it doesn’t raise your rankings on its own, and Google has made it clear that it isn’t a direct ranking factor. It does, however, make the page easier to understand and can trigger “rich” results, such as review stars and prices. On top of that, Microsoft has confirmed that schema helps Bing’s and Copilot’s AI models understand content.
Entities: from keywords to “things”
Since 2012, Google has described its shift with the phrase “things, not strings.” It doesn’t just look for words; it tries to understand “things,” in other words, entities. An entity is your business as something unique, with a name, address, founders, services and relationships with other entities. AI tools “think” in a similar way.
To make it clear who you are:
- Same details everywhere: The same name, address and phone number on your site, on your Google Business Profile and in directories.
- Linking your profiles: In your schema, the sameAs field connects your site to your official profiles (LinkedIn, Facebook, Instagram), making it clear that they all belong to the same business.
- Real details on your “About” page: Year founded, team, certifications. Anything that can be verified.
The clearer your “entity,” the easier it is for ChatGPT, too, to recommend you correctly, as we explain in our guide to GEO.
Check your site in 15 minutes
- The Ctrl+U test: Is your text in the source code?
- Rich Results Test and Schema Markup Validator: Run a page through Google’s tool and validator.schema.org and see whether it has structured data and whether there are any errors.
- Search Console: In the URL Inspection tool, see the page the way Google “sees” it.
- Search for your name: Does Google show a panel with your business details? Are they correct?
Why a plugin won’t fix it
Many sites add schema with a plugin and consider the job done. In practice, plugins add generic information, often duplicated or contradictory. And they can’t fix a site that loads its content with JavaScript or has “bloated” code from a page builder.
Structure, clean HTML and structured data work best when they’re designed together with the site itself: each page type (service, product, article) automatically generates the right schema, and the content arrives ready-made in the code that bots read. That’s why technical SEO is now part of web development, not something added at the end.
