Which content on your site is invisible to AI retrieval?
A fact your customers can read is not automatically a fact a machine can read, and the gap is usually in six predictable places.
The sentence on your homepage that says you see new patients on Saturdays may exist only for people using a browser, and not for the software deciding whether to mention you. Nothing looks broken. The page renders, the phone rings, and yet a crawler receives a page with that sentence missing.
This is where a lot of lost AI discoverability hides on otherwise good small business sites. Below are the six usual places, each with a test you can run yourself without any tools beyond a browser. For the underlying idea, see the lexicon entry on retrieval legibility.
Six places facts go missing, and how to test each
1. Text that only appears after JavaScript runs
Many site builders and themes load text into the page with JavaScript after it arrives. Google renders pages with a headless Chromium browser, per its JavaScript SEO basics, but the same page says server side rendering or pre-rendering is "still a great idea" because "not all bots can run JavaScript." Treat any key fact that depends on JavaScript as a fact some AI crawlers will not receive.
Test: open the page, right click, choose View Page Source, and use find to search for a sentence you care about, such as your hours. If it is in the source, it arrives with the page. If it is not, it is being added later.
2. Loading placeholders
A close cousin of the first problem. Some pages ship a spinner or the word "Loading" in place of the content, then fetch the real text. A reader that does not run the script gets the placeholder and nothing else. For a booking widget that is minor. For a page whose entire body loads this way, the page is effectively empty to that reader.
Test: in View Page Source, search for "Loading" or look at how much readable text there is at all. A source file that is almost all script tags and very little prose is the warning sign.
3. Facts written inside images
The banner that lists "Family, cosmetic and emergency dentistry, Ventura and Oxnard" is often a single picture. Google's image guidance calls alt text "the most important attribute" for describing an image, so the words a machine can count on are the alt text and the HTML around the picture, not the lettering in it.
Test: try to highlight the words with your mouse. If you cannot select individual words, they are part of an image. Put the same words in real text on the page, and give the image honest alt text.
4. The PDF menu, service list or intake form
Restaurants, clinics and trades often keep their most specific information in a PDF. Google lists PDF among the file types it can index, so the file is not hopeless. But it is a separate document, often with no business name or city in the text, and any passage quoted from it arrives without the context of your site around it. A scanned PDF that is only a picture of a page carries the image problem as well.
Test: open the PDF and try to select a line of text. Then ask whether the key items (the dishes, the services, the areas served) also appear on an HTML page. If they only live in the PDF, copy the essentials onto a page.
5. Labels that look like headings but are not
A section titled "Emergency appointments" in large bold type may be marked up as an ordinary paragraph rather than a heading. People see a heading. Software that divides a page into sections at its headings sees one long run of text with no label, and the section loses the words that said what it was about. The lexicon entry on chunking explains why those section boundaries matter.
Test: right click the label and choose Inspect. If the highlighted tag is h2 or h3, it is a heading. If it is p, div or span styled to look large, ask your web person to change the tag, not the look.
6. Pop-ups that cover the page
A full screen coupon or newsletter overlay is a problem for more than visitors. Google's guidance on intrusive interstitials says such dialogs "make it hard for Google and other search engines to understand your content," and recommends small banners instead.
Test: load the site on your phone in a private window. If the first thing you see is something you must close before you can read anything, replace it with a banner.
What to fix first
Start with the facts a customer would ask an assistant about: what you do, where, for whom, when you are open, and how to reach you. Run the View Page Source test on the homepage and the main service page for just those facts. If they are in the source, in real text, under real headings, you have covered most of the risk. The remaining items matter in proportion to how much of your specific information lives in them.
Two neighboring questions are worth checking next. Content is only readable after a crawler is let in, so confirm that first with the page on whether AI crawlers can read your site. And once the text is visible, the most important piece of it is a single sentence, covered in name, city and service in one sentence.
Hidden content is one of the more fixable parts of AI discoverability, because every item here has a yes or no test. The AI discoverability guide puts it alongside the other signals, and you can run the free AIOInsights check for a reading of your own homepage.
Questions
How can a business owner tell if AI crawlers can see the text on their website?
A business owner can open the page, choose View Page Source in the browser, and search the source for a key sentence such as the opening hours or the phone number. If the sentence is in the source, a crawler that does not run JavaScript receives it. If it only appears after the page loads, some AI crawlers may never see it, because Google notes that not all bots can run JavaScript.
Is text inside images readable by AI search?
Text baked into an image, such as a banner that lists services or a graphic with opening hours, should be treated as unreliable for AI search. Google calls alt text the most important attribute for describing an image, which means the words a machine is sure to receive are the alt text and the surrounding HTML. Put every important fact in real page text as well.
More on AI discoverability
This page is part of AI Discoverability, the AIOInsights guide to whether AI systems can find, read and name a business.
- How do AI assistants find a business to name in an answer?
- Can the AI systems that answer questions actually fetch your pages?
- Why one sentence with your name, city and service does so much work
- What happens when your business details disagree across the web?
- The sites AI reads about you, and why your own site is often not the one it cites
- Does structured data help AI find your business?
- How to measure AI discoverability without fooling yourself
- A first week plan for AI discoverability
- Where AI discoverability work falls short