Login Walls, Paywalls and Cookie Banners: What AI Assistants See

Anything sitting behind one of those doesn't exist for an assistant. Not partially, not in summary. Not at all.

The page still looks normal to you, because your browser remembers that you're logged in and that you accepted the banner months ago. Open the same page in a private window and you'll see what an assistant sees.

• A login wall: the page redirects anyone who isn't signed in. Common on membership sites, client portals and sites where a plugin was set to "force login" site-wide by mistake.
• A paywall: part or all of the content is reserved for subscribers.
• A consent wall: a cookie or privacy banner that covers the page and loads nothing until someone clicks. A banner you can scroll past is fine. One that holds the content hostage is a wall.

The third one surprises people most, because it was installed for legal compliance and nobody thought of it as blocking anything.

Walls are often applied far wider than intended. A membership plugin set to protect one area ends up protecting the whole site. A consent tool defaults to blocking everything until a choice is made. Nothing warns you, because you're always on the inside.

1. Open your site in a private browser window. No login, no history, no saved consent. This is the single most useful check in this whole article: it shows you the page as a stranger, and as an assistant, receives it.

2. Decide what genuinely needs protecting. Prices, services, locations, about and contact pages almost never do. Client areas and paid courses do.

3. Move what should be public outside the wall. In most plugins this means narrowing protection from "whole site" to specific pages, rather than turning the plugin off.

4. Make your consent banner non-blocking. The page content should load and be present in the page source whether or not anyone clicks. Most consent tools have a setting for this; it's the difference between a banner and a gate.

5. If you run a real paywall, mark it up properly. Google supports structured data for paywalled content and describes acceptable approaches in its flexible sampling guidelines. This is the supported way to have both a paywall and visibility.

One thing not to do: detect crawlers and serve them the full content while people see a wall. That's cloaking, it's against search guidelines, and the penalty is worse than the problem.

Some security and bot-protection services show a challenge page ("checking your browser") to visitors they don't recognise. People pass it in a second and never notice. Crawlers often don't pass it at all. If your private-window test looks fine but assistants still see nothing, that's the place to look.