Paywall SEO: how to rank content behind a login wall

Googlebot does not log in, subscribe or pay. Left alone, it sees whatever a logged-out visitor sees: a preview and a wall, or a redirect to a login page, and it indexes that. For most gated content that is the right outcome, because most gated content is post-purchase material (reports, onboarding, account pages) that nobody searches for before buying, and our gated content strategy guide covers why it belongs behind the wall. But some gated content is the best answer on the web to a question prospects ask: the members' version of a guide, the paid course's opening lesson, the subscriber-only analysis. For those pages Google offers a supported path, documented under structured data for subscription and paywalled content, that lets the whole article be indexed while the reader still meets the wall.

The path has three parts, all on the article's own page. The article's structured data carries isAccessibleForFree set to false, which tells Google the page is intentionally partly gated. A hasPart property holds a WebPageElement, also with isAccessibleForFree false, whose cssSelector names the class of the gated section. And in the HTML, the gated section is wrapped in an element carrying that class, with the full text served to Googlebot inside it. Google's documentation is specific on the details: class selectors only (never an id or a nested selector), sections not nested inside one another, and JSON-LD or microdata for the markup. Google's guidance applies the same way to a metered paywall, a hard paywall or a login wall, because from the crawler's side they are the same thing: content it can read that a visitor sometimes cannot.

Cloaking is showing search engines different content from what a person sees, and it is a violation of Google's spam policies that can remove a site from results. A paywall does exactly that: Googlebot sees the full article, a logged-out person sees a preview. What makes one a supported practice and the other a policy violation is the declaration. Google's own documentation says its crawlers need to see the full content to index it, and that marking the gated section with the structured data above is how you distinguish paywalled content from cloaking. The markup is not an optimisation. It is the licence.

Two failure modes follow from that. A site that serves Googlebot the full article and skips the markup is, by Google's definition, cloaking, however innocent the intent. And a site that adds the markup but serves Googlebot the same truncated preview a person sees has declared a wall that hides nothing, so the gated body is never indexed and the markup does nothing. The crawler must get the full text and the declaration together, on the same page, every time.

In October 2017 Google retired its First Click Free policy, which had required publishers to show a visitor arriving from search the full article before asking them to subscribe, and replaced it with flexible sampling. Under flexible sampling the publisher decides how much a non-subscriber sees, in one of two ways: metering, where a visitor gets a number of free articles per month before the wall appears, or lead-in, where every article shows a portion of its text before the wall. Google's guidance at the time suggested publishers start around ten free articles a month under metering and adjust from there, and it made clear that a site was free to sample less, with the structured data above doing the work of keeping the page indexable regardless.

For a business site rather than a news publisher, lead-in is nearly always the right model. A members' guide shows its answer-first opening and its key figures to everyone and gates the depth; a course shows the framework and the first lesson. That lead-in is also the only part of the page an AI engine will ever quote, which makes its length and content a strategic decision rather than a technical one.

The distinction that matters is not how the visitor pays but whether a stranger would search for the page before buying. The rule from our search intent guide holds: open before the sale, gated after it. Paywalls and free login walls both sit on pre-purchase content that deserves to be found, so they get the full text plus the markup. Hard gates sit on post-purchase content that belongs to a customer, so they are excluded cleanly, with a noindex or a login redirect rather than a soft 404 or an infinite redirect that wastes crawl budget. The client portal guide covers where each kind of page lives in the two-layer site.

ChatGPT, Google AI Overviews and AI Mode, and Perplexity read what a crawler can read, and they cite what they can quote. For a page with the paywall markup, Google's systems have the full text and can use it to judge relevance, but the answer an engine shows a user is built from what the user could also see, which is the preview. In practice that means a gated guide is cited for its lead-in or not at all. The strategic consequence is that the open portion of a gated page has to carry the definition, the key figure and the named source you want attributed to you, and the gated portion holds the depth the reader logs in for. Written that way, the paywall costs nothing in citations that the page would otherwise have earned, because the quotable part was always going to be the opening.

Two things not to do. Do not use the markup to hide an entire article behind a one-line preview and expect it to rank or be cited; a lead-in that answers nothing earns nothing. And do not rely on the markup for content that should simply be open; if a prospect would search for it before choosing you, gating it costs the ranking and the citation whatever the schema says. Our guide to getting cited by AI in Australia covers what the open portion has to look like.

Gated section wrapped in a single class; isAccessibleForFree false on the article node; hasPart WebPageElement with the same class in cssSelector; class selectors only, nothing nested; Googlebot served the full text on the same URL; the boundary identical for every logged-out visitor; a lead-in that states the answer and the key figure; post-purchase pages excluded with noindex or a login redirect, never a soft 404; and the page verified in the Rich Results Test and fetched as Googlebot in Search Console. Nine checks, an afternoon's work for a developer, and the difference between a members' library that ranks and one that is invisible.

Auditing a gated site against those nine is part of every SEO engagement we run, and our Snowball SEO platform monitors the indexation and citation of the open previews afterwards, which is the ongoing checking other agencies bill by the hour. On this site the value is true everywhere because the hub is open; the schema is already built so that flipping one page to false is a one-line change, which is the right way to build a site that might one day gate part of itself.