Most sites that are invisible in AI answers are not being punished. They are being skipped, because the thing a buyer wanted to know was baked into a picture, folded inside an accordion, or loaded by a script after the page arrived. Google's documentation asks for the opposite in plain language. This page helps you find what your own site is hiding and get it back into text.
Source: Google Search Central, "AI features and your website", checked October 2026.
Where to start
The two-minute test
Open your most important page. Press Ctrl+U on Windows or Cmd+Option+U on a Mac to view the source — the raw HTML the server sent, before any script ran. Now use your browser's find to search that source for a distinctive phrase from your key answer.
If the phrase is there, you are fine. If it is not, something between your writer and your reader has swallowed it, and anything reading the page without running your scripts sees the same blank space you just found.
Google's guidance on this is direct. Its page on AI features lists keeping important content in text form among the things that matter, and adds that "You don't need to create new machine readable files, AI text files, or markup" to appear in these features. It also states that "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary."
Read those together and the instruction is unusually clear: stop looking for a special file to add, and check that your actual answer is actual text.
The three hiding places, and what to do about each
Inside an image. Pricing tables, comparison charts, spec sheets and infographics are the usual offenders. A designer exports a beautiful table as a PNG and the numbers leave the web. The fix is to build the table in HTML and keep the picture as decoration, or drop the picture. Alt text is not a substitute — it is a short description, not a place to hide a table.
Behind a click. Accordions, tabs and "read more" toggles. These vary. If the text is in the HTML and merely hidden with CSS until someone clicks, it is usually readable. If the click is what fetches the text, it is not there at all until a human acts, and no crawler acts. The source-view test tells you which kind you have.
Rendered by script. Single-page apps and heavy frameworks often ship a near-empty HTML shell and build the page in the browser. Google can render JavaScript, but rendering is a second pass that is not guaranteed for every page every time, and several other AI crawlers do far less of it. If your answer only exists after JavaScript runs, you are relying on the most fragile path.
Find what is hidden
Answer for one page. Nothing is sent anywhere.
See the page the way a machine does
Two checks beyond view-source, both free.
Turn JavaScript off and reload. In Chrome, open Settings, search for JavaScript, and block it for the site. What remains on screen is roughly the worst-case view a simple crawler gets. If the page is blank, you have learned something important.
Use Search Console's URL Inspection. For any page on a property you own, "View crawled page" shows you the HTML Google actually got. This is the authoritative version for Google specifically, and it beats guessing.
When you find a gap, resist the urge to rebuild the whole site. Put the answer in text near the top of the page, as ordinary paragraphs under a heading that matches the question. That single change fixes most cases.
Rewrite helper
Turn a hidden answer into a block you can paste into the page.
- View-source your five most important pages and search for the key answer in each.
- Rebuild any pricing or comparison table that currently exists only as an image.
- Check whether your tabs and accordions ship their text or fetch it. Fix the ones that fetch.
- Reload one page with JavaScript blocked and look at what survives.
- Put the answer in plain paragraphs near the top, under a heading in the buyer's words.
- Do not add a special AI file. Google says in writing you do not need one.
Where this guidance stops.
Google can render JavaScript and often does, so "my site is a single-page app" is not an automatic death sentence in Google. The argument for plain text is that it is the only approach that works everywhere without depending on a rendering pass you cannot observe or control.
Other engines publish much less. OpenAI, Perplexity and Anthropic do not document their rendering behaviour in the detail Google does, so the honest position is that nobody outside those companies knows exactly how much JavaScript each one executes. That uncertainty is itself the argument: text costs you nothing and removes the question.
Finally, being readable is necessary, not sufficient. A page an engine can read perfectly and still finds unhelpful will not be cited. This fixes the floor, not the ceiling.
Sources: Google Search Central, "AI features and your website" (checked October 2026). Quotations verbatim from that page. Every tool here runs in your browser; nothing is sent anywhere.