Can ChatGPT read your website?
Content AI cannot read: cookie walls, logins, PDFs and text in images
Your website can be open to crawlers and still hide your key facts: behind a cookie wall, a login, a scanned PDF or an image. In an hour, four simple checks show where your opening hours, prices and services are hiding, and how to move them back into text on the page.
· By the Letzscale team · 9 min read
On this page
- Time
- 1 hour
- Difficulty
- Easy
- Who
- You, with whoever edits your website
- You need
- Google Chrome on a computer and a list of your key facts
- Checked
- 24 October 2026
Part of the Letzscale guides: free, step-by-step fixes that help ChatGPT and other AI assistants find, understand and recommend your business. New here? Start with Guide 01: Check what ChatGPT says about your business.
Why it matters
Your website can be open to crawlers and still hide the facts that matter. A crawler is an automated visitor: it has no password, it does not click "Accept", and it reads text, not pictures. When your prices, opening hours or services sit behind a cookie wall, behind a login, inside a scanned PDF or inside an image, a crawler may find a page with almost nothing to quote.
Google's advice for its AI features is plain: make sure "important content is available in textual form", and use images and videos to support that text, not replace it. This guide finds the facts that are hidden on your website and moves them into normal page text.
<h1>Café Ribeiro</h1> <p>Discover our menu!</p> <a href="https://cafe-ribeiro.example/menu-october.pdf"> Download the menu (PDF)</a> <!-- the PDF is a scan: a photo of the printed menu -->
<h1>Café Ribeiro: lunch menu</h1> <h2>Lunch, Monday to Saturday, 12:00 to 14:30</h2> <p>Dish of the day: €16.50</p> <p>Soup of the day: €7.00</p> <p>Pastel de nata: €2.80</p> <a href="https://cafe-ribeiro.example/menu-october.pdf"> Menu to print (PDF)</a>
Illustration with a fictional business.
Reference
Four places facts hide
| Where | What a crawler gets | The fix |
|---|---|---|
| Cookie wall The page stays blank or sends you to a consent page until you click "Accept" | Possibly only the consent page. Google says redirecting visitors to a separate consent page leaves only that one page for Googlebot to fetch. | Keep the consent banner, but show it on top of the page, with the content already loaded underneath. |
| Login Customer area, members' prices, client portal | Nothing. Google: "If a page is made private, such as requiring a log-in to view it, Googlebot will not crawl it." | Put the public facts (services, prices from, hours, contact) on public pages. Keep only customer-specific content behind the login. |
| PDFs Menus, price lists, brochures | Google can index PDFs, but a scanned PDF is a picture with no text in it, and the PDF is a separate document, not your page. | Key facts as text on the page, the PDF as an extra download. |
| Images, sliders and videos Opening hours on a poster image, prices in a slider, a video tour | A picture. Google reads the alt text and nearby text to understand an image. | Real text for important information, alt text for images that carry information, a text description or transcript for videos. |
Before you start
Have these ready
- Google Chrome on a computer. The shortcuts below are Chrome's own.
- Your key facts: opening hours, prices or "from" prices, services, address, phone. If you have a fact sheet from Guide 04, use it.
- The three to five pages where these facts should be: usually the homepage, the services or menu page, the prices page and the contact page.
Steps
Four checks, one per hiding place
Check 1: open your pages as a first-time visitor
Open a private window: Ctrl + Shift + N (Windows, Linux) or ⌘ + Shift + N (Mac). Chrome calls it an Incognito window. It has no saved cookies and no logins, which is closer to what a crawler sees. Open each of your pages in it.
- A cookie banner appears over the page, and you can see your text behind or below it. Good: the content is on the page.
- The page stays empty, or your address changes to a separate consent page, until you click "Accept". That is a cookie wall. See fix A.
- The page asks you to log in. Anything on it is invisible to crawlers. See fix B.
Check 2: search the page source for your facts
On each page, press Ctrl + U (Windows, Linux) or ⌘ + Option + U (Mac). Chrome shows the HTML file exactly as your server sent it. Press Ctrl + F or ⌘ + F and search for a price, for example
16.50, and for your opening hours, for example14:30.- Found. Good: the fact is text on the page.
- Not found, but you can see it on screen. It is in a PDF, an image, or built by JavaScript. Checks 3 and 4 tell you which. To test for JavaScript, run the "Disable JavaScript" check in Guide 12.
Check 3: test your PDFs
Open each PDF that holds a key fact (menu, price list, brochure) in Chrome. Press Ctrl + F or ⌘ + F and search for a price. Then try to select a line of text with your mouse.
- Found, and you can select the text. The PDF has real text. Google can index PDFs, but the fact still lives in a separate file, not on your page. Copy the key facts onto the page too (fix C).
- Not found, or you cannot select anything. The PDF is a scan or a picture: there is no text in it to read. Fix C is a priority.
Check 4: look for text inside images, sliders and videos
On each page, try to select the important words with your mouse: opening hours, prices, offers, phone number. Words you cannot select are part of an image. Look especially at banners, posters, sliders and "special offer" graphics.
In the page source (check 2), search for
alt=. Each image has analtattribute: a short text that describes the image. If an image that carries information has an emptyalt=""or none at all, see fix D.For videos, check that the page around the video says in words what the video says: what you offer, where, and any prices or dates it mentions.
Fix
Move the facts into page text
Fix A: a cookie banner that does not hide the page
Keep asking for consent. Change how the banner is shown: the page content should load as normal, with the banner on top of it, not instead of it. Google's guidance for consent and other required pop-ups is to make sure "the content is overlaid with the interstitial", and not to redirect visitors "to a different page for collecting consent". If your site sends everyone to one consent page first, Google can only fetch that one page.
Most websites use a cookie tool or plugin for the banner. Ask your developer or the tool's support for a banner that sits over the page, with no redirect to a separate consent page and no blank page before the choice.
In Luxembourg: keep the banner, do not lock the page
The fix is not to remove consent. Cookies that are not strictly necessary, for example for analytics or advertising, still need visitors' consent. The CNPD, Luxembourg's data protection authority, publishes guidelines on cookies and other trackers. They say a site that makes access to its content depend on clicking "I accept" does not collect valid consent, and they recommend that refusing is as easy as accepting. Read the CNPD guidelines on cookies (in French) and check your banner with whoever handles data protection for you. This guide is not legal advice.
Fix B: public facts on public pages
Go through what sits behind your login. Anything a new customer needs to choose you belongs on a public page: what you offer, for whom, your prices or "from" prices, your hours, how to contact you. Keep only customer-specific content behind the login, such as documents, orders and personal prices.
If you charge different prices to members, a public page can still say what the service is and how to become a member.
Fix C: key facts as page text, the PDF as an extra
Type the important parts of each PDF into a normal page: dishes and prices, service names and prices, opening hours. Use headings and short lines, as in the Café Ribeiro example above. Keep the PDF for printing, with a link that says what it is, for example "Menu to print (PDF)".
Change both at the same time. A page that says €16.50 and a PDF that still says €15.00 gives people and AI assistants two answers. A line such as "Prices valid from 1 October 2026" on the page helps.
For how to write the page itself, so the first lines answer the questions people ask, see Guide 06: Write pages ChatGPT can quote.
Fix D: real text instead of text in images
Where an image shows words that matter (opening hours, a price, an offer), put those words in a normal text block on the page. The W3C, which sets the web's accessibility standards, calls it good practice to use real text styled with design settings rather than text in images.
Give every image that carries information an
alttext. Google uses alt text, together with what it sees in the image, to understand the image, and alt text helps people who use screen readers. If the image shows text, the W3C says the alt text "must contain the same text presented in the image". Google's own example of a good alt text is short and specific: "Dalmatian puppy playing fetch", not a list of keywords.Image alt text developer<img src="opening-hours.jpg"> <img src="opening-hours.jpg" alt="Opening hours: Monday to Saturday, 8:00 to 18:00. Closed on Sundays.">In WordPress, Wix and Webflow, you add alt text in the image settings, usually in a field called "Alt text". Even with good alt text, keep important facts in the page text too.
For videos: give the page a title and a description that say what the video is about. Google asks for a title and description that are unique to each video page. Add a transcript, a text version of what is said in the video, below it. The W3C notes that a transcript with headings and links "also helps with SEO".
Check it worked
Run the checks again
After the changes go live, repeat checks 1 and 2 in a new Incognito window, with the same pages and the same facts.
Done when
- In an Incognito window, every page shows its content, with the cookie banner on top and no redirect to a consent page.
- Your opening hours, prices and services appear in the page source of public pages.
- Every PDF with key facts has a page that says the same thing in text.
- Images that carry information have alt text, and the same facts are in the page text.
Common mistakes
What usually goes wrong
- Removing the cookie banner. The problem is a banner that blocks the page, not the banner itself. Keep asking for consent.
- Testing in your normal browser window. You have already accepted cookies and you may be logged in, so you do not see what a first-time visitor sees.
- Updating the PDF and forgetting the page, or the other way round. Two different prices are worse than one.
- Stuffing alt text with keywords. Google warns this can make your site look like spam. Describe what the image shows.
- Opening hours only in a slider or a poster image. The most asked question about your business deserves a line of plain text on your homepage and contact page.
Sources, checked 24 October 2026
- Google: AI features and your website
- Google: File types indexable by Google
- Google: Google Search technical requirements
- Google: Interstitials and dialogs
- Google: Image SEO best practices
- Google: Video SEO best practices
- W3C WAI: Images of text
- W3C WAI: Transcripts
- CNPD: Lignes directrices en matière de cookies et autres traceurs · 3. Principes applicables et mise en pratique (in French)
- Google Chrome Help: Browse in Incognito mode
- Google Chrome Help: Chrome keyboard shortcuts