Machine-readable content
Machine-readable content is text a bot can read and understand without guessing: in the page's HTML, in clear structure, with no key facts hidden in images.
Machine-readable content is content that software can read and make sense of without guessing. For a web page, that means the important text is in the HTML your server sends, laid out with real headings, lists and tables, and not trapped inside images, scripts or tabs.
AI platforms can only use what their bots can read. A page that looks perfect to a person can look almost empty to a bot.
What bots miss
Most AI bots read the HTML your server sends and stop there. They don't click, scroll or wait for scripts. So they can miss:
- Text that a script loads after the page opens
- Text inside images, like a price on a banner
- Content behind tabs or "read more" buttons that only loads when clicked
- Tables built from images or scripts instead of HTML
For example, Harrow & Pine's menu page shows prices in a slick slider that a script builds after the page loads. People see every loaf and its price. A bot sees the heading "Our bread" and nothing else, so AI answers about Harrow & Pine's prices quote an old blog post instead.
How to make content readable
Put the text in the HTML. Use one main heading, then headings for each section in order. Use real lists and tables for lists and tables. Give images alt text that says what's in them. Add structured data, such as Product or Organization markup, so key facts are labelled.
A quick test: turn off JavaScript in your browser, or view the page's source, and see what's left.
In AnswerOrbit
The Pages tab in Website shows each page side by side as What a person sees and What a bot reads. Blocks bots miss are marked Hidden from bots, with why and how to fix it. See Website and AI traffic.