Your page looks fine on your phone. AI might not be able to read it

The screen is the finished dish. The raw HTML is the ingredients that arrived first

When you open a page, the browser first receives an HTML document from the server. Then it runs the JavaScript linked from that document to load the price, attach reviews and fill in the product details. What you see on your phone is the result after all of that has finished.

Many stores and websites today fill in a lot of content during that second step. People don't notice the difference, because it takes less than a second. The problem is anything that reads only the first document and never runs the JavaScript. From there, the price slot and the review slot look empty.

One product page, two versions Example

Your phone screen After JavaScript runs Product name · $29 "Waterproof fabric, carry-on size" 128 reviews · shipping info This is what people see.

The raw HTML first received Before JavaScript runs Product name <div id="price"></div> empty <div id="detail"></div> empty Anything reading only this stops here.

Not every site works this way. Plenty of sites send the content in the HTML from the start. The 3-minute check below tells you which kind your page is.

Which version do AI crawlers read?

Google explains that it processes JavaScript pages in three phases: crawling, rendering and indexing. In the rendering phase, Google Search runs the JavaScript and sees the finished page. Yet the same document still recommends server-side rendering or pre-rendering, because it makes pages faster for both users and crawlers, and because not all bots can run JavaScript.

What about AI crawlers? In December 2024, Vercel published an analysis of crawler requests on its network, done with MERJ. The findings: none of the major AI crawlers from OpenAI, Anthropic, Perplexity and others rendered JavaScript. ChatGPT's and Claude's crawlers did download JavaScript files (11.50% and 23.84% of their requests) but did not execute them. Gemini, on the other hand, uses Googlebot's infrastructure and can render, and AppleBot rendered through a browser-based crawler.

What each crawler takes away (Vercel and MERJ analysis, December 2024)

ChatGPT (OpenAI)Raw HTML onlyDownloads JavaScript files but did not execute them.

Claude (Anthropic)Raw HTML onlyDownloads JavaScript files but did not execute them.

PerplexityRaw HTML onlyDid not render JavaScript.

Gemini · GoogleThe finished pageRuns JavaScript on Googlebot's infrastructure. AppleBot also rendered.

This is a public analysis from December 2024. Crawler behavior can change, so take away the principle rather than the snapshot: if the key content is in the raw HTML, you're covered either way.

Vercel's recommendation points the same way. Render critical content on the server so it arrives in the first HTML, and leave only non-essential parts, like view counters or live chat widgets, to the browser. The price, specs, shipping and exchange terms a buyer checks before purchasing are clearly critical content.

What tends to go missing from the raw HTML

It depends on how the site was built, but when I go through a page with someone, the places worth checking are fairly predictable. Most of them are content that gets "loaded later".

Places to check first in the raw HTML Example

Product nameUsually in the raw HTML, often in the title tag as well.
Price · optionsBecause they change with stock and discounts, some sites load them after the page opens.
Details · reviewsContent behind "see more", infinite scroll or a separate widget may not be in the raw HTML.
Shipping · returnsIf your site loads these as a shared block, check them too.

Red doesn't mean it's definitely missing. Every site is different, so opening the raw HTML yourself is the surest way to know.

One more thing. You don't need to rebuild your site if something is missing. Start by asking your platform or developer: "Can the price and product details on this page come out of the server as HTML from the start?" Many already support it under names like server-side rendering or pre-rendering. Having content in the raw HTML still doesn't guarantee an AI will cite your page. Making it readable is the part that's in your hands.

A 3-minute check you can do today

On a computer, open the product page you most want to sell. It's much easier on a PC than on a phone.

1 min · OpenOpen the page sourceIn Chrome or Edge, press Ctrl+U (Cmd+Option+U on a Mac). The raw HTML the server first sent opens in a new tab.
1 min · SearchSearch for three key linesWith Ctrl+F, search for the price, one sentence from the product details, and one sentence from your exchange terms.
1 min · NoteWrite O or XFound it: O. Not found: X. If there's even one X, you have your first question for your developer.

Where doing this by hand gets stuck

One page takes three minutes. But with dozens of products, several templates, and the question of how the sentences in that raw HTML actually read to shoppers and AI, you end up repeating the same work page after page. And once developer jargon creeps in, it's easy to put off.

AIOudit takes that part off your hands. Paste a page URL and it looks at how shoppers and AI read that page in one pass, then lays out what to fix first.

AIOudit can't see your price competitiveness, stock, ads, seasonality or brand reputation. And which page an AI cites is ultimately up to each AI.

See how shoppers and AI read your page. Run it through AIOudit.