Machine Readability: How Machines Read Your Pages

Your Page Has Two Readers

When a person visits your page, they see the finished product: typography, images, color, layout. When a machine visits — Google’s crawler, Bing’s, a social network building a link preview, or an AI engine deciding whom to cite — it sees none of that. It sees markup. Tags, attributes, and structured statements about what the page is.

Most sites are written entirely for the first reader. The second reader is left to guess. And a machine that has to guess does something worse than misunderstand you — it hedges. It ranks you a little lower, previews you a little plainer, cites you a little less often than the competitor who spelled things out.

Machine readability is the practice of spelling things out. It has three layers, and you can audit all three in about ten minutes.

Layer One: Can Machines See Your Content at All?

Right-click any page and choose “View Page Source.” What you’re looking at is the raw HTML — the version of your page every machine receives first. Now search that source for a sentence from the middle of your page. If you find it, your content lives in the HTML. If you don’t, your content is being assembled in the visitor’s browser by JavaScript — and some of your most important readers never run it.

Google can render JavaScript, but rendering is expensive, so it happens on a delay — sometimes days — and not always completely. Many AI crawlers and preview builders don’t render it at all: what’s not in the raw HTML simply doesn’t exist for them. A beautiful single-page app can be, to a machine, an empty room.

The fix isn’t abandoning JavaScript — it’s making sure the content that matters (headlines, body copy, titles, descriptions) is server-rendered into the HTML, with JavaScript adding behavior on top. If your platform can’t do that, treat it as a real cost when you weigh what that platform is worth.

Layer Two: Open Graph — the Clothes Your Link Wears

Every time your page is shared — in a text, a Slack channel, a LinkedIn post, or an AI chat answer — the preview card is built from a handful of tags in your page’s head: og:title, og:description, and og:image. Pages without them get the plain-text treatment: a bare URL, maybe a scraped fragment, no image. Pages with them get the full card.

This is the cheapest professional dress code on the web. Three tags per page, set once. The difference in click-through between a bare link and a rich card is not subtle — and increasingly those cards appear in places you don’t control, including AI answers that link out to sources.

Layer Three: Schema — Introducing Yourself in a Language Machines Trust

Schema markup (structured data, usually written as JSON-LD) is a small script in your page that states plainly what the page is: this is an article, written by this person, published on this date. This is a business, at this address, with this phone number. These are frequently asked questions, and here are the answers.

Without it, machines infer those facts from context, with the confidence of a stranger reading over your shoulder. With it, they stop inferring and start knowing. That knowledge shows up two ways:

Rich results. Review stars, FAQ dropdowns, breadcrumbs, sitelinks, event cards — the search features that make a listing physically larger and more clickable are almost all schema-fed. A page can rank well and still lose the click to a schema-dressed competitor below it.

Entity clarity. Search and AI engines model the world as entities — this business, this person, this organization — and connect what they read to those entities. Consistent schema (matching your name, address, and phone everywhere they appear) is how a machine becomes certain that the firm on your homepage, the profile on a directory, and the author of an article are the same trusted thing. Certainty is what gets cited.

What AI Engines Do With All of This

When an AI answer engine composes a response, it isn’t ranking ten blue links — it’s choosing a handful of sources to quote and cite. Watch what wins those citations: content that was present in the raw HTML, described accurately by its tags, and attributable to a clearly identified entity. In other words, the three layers above aren’t three separate chores. They’re one property: being quotable.

The old game was persuading an algorithm to rank you. The new game — layered on top, not replacing it — is making it effortless for a machine to understand you, preview you, and vouch for you. Sites that are easy to quote get quoted.

The Ten-Minute Machine Readability Check

For any important page:

Or let the tools do it: our free Schema Checker reads your structured data exactly the way machines do — every block parsed and graded — and the SEO URL Checker covers rendering and Open Graph alongside the fundamentals from Chapter 1.

The Bottom Line

You already write for the reader you can see. The rankings, the rich previews, and the AI citations belong to sites that also write for the reader they can’t. Say what your pages are, in the languages machines read — and both audiences will find you.