96.3% of New York Times citations in AI Overviews come from content behind a paywall. The Washington Post figure is 99.13%. These numbers do not mean paywalls help AI citation – they reflect that Google’s crawler has special access to paywalled content from major publishers via Flexible Sampling, while simultaneously those publishers put nearly all content behind paywalls. The mechanism matters more than the surface statistic.
Google’s Stated Policy on Paywalled Sources in AI Overviews
Google’s official position from Google Search Central documentation updated December 10, 2025: AI Overviews and AI Mode are subject to the same Search preview controls as traditional results. For paywalled content to be eligible for AI Overview citation, Googlebot must be granted access to crawl and index the paywalled sections. Publishers who want paywalled content included must allow Googlebot and Googlebot-News through their access controls.
The technical implementation requires schema.org JSON-LD with isAccessibleForFree: false in NewsArticle structured data, with paywalled sections identified by CSS class selector. This schema setup tells Google that the content exists and is gated – it does not automatically grant AI Overview citation eligibility, it signals the access model so Google can process it correctly.
The Flexible Sampling mechanism is the operational path for major publishers. Publishers who grant Googlebot access to paywalled content allow that content to be indexed and cited in AI Overviews even though end-users cannot access it for free. This creates the observable paradox: 96% of cited NYT content is paywalled, yet citations appear in AI Overviews. Google crawled it with publisher permission, indexed it, and now cites it while the reader clicking the citation link hits a paywall.
The Practical Difference in Citation Rate Between Open and Gated Content
Hard paywall without Googlebot access eliminates AI Overview eligibility entirely. If a publisher blocks Googlebot from paywalled sections – or uses a client-side JavaScript paywall that Googlebot cannot bypass – those pages are not indexed and therefore cannot be cited in AI Overviews. Hard paywall with zero Googlebot access is equivalent to a noindex directive for AI Overview purposes.
Common Crawl research found that many legacy paywalls fail to block bot crawlers at the HTML rendering level – bots receive full content before JavaScript executes the paywall. This is a technical implementation gap, not an intentional access grant. Publishers who believe they have a hard paywall should verify bot access behavior using Screaming Frog’s custom user agent configuration or Google Search Console’s URL Inspection tool requesting as Googlebot.
Open access content has no crawl barrier and eliminates one friction point in the citation selection process. All else equal, an open access page and a paywalled page of equivalent quality and authority compete on the same signals – but the paywalled page requires the publisher to have explicitly granted Googlebot access via Flexible Sampling. Open access is the simpler path to AI Overview eligibility for publishers who do not have Flexible Sampling configured.
When Metered Paywalls Allow Enough Crawl Access for AI Overview Inclusion and When They Block It
Metered paywall access model: publishers who allow a defined number of free article views per month – typically one to three articles – maintain AI Overview eligibility because the first access to each page is open before the meter triggers. This is the most AI-Overview-compatible access model for publishers trying to balance citation visibility with subscription revenue.
Content behind a meter is crawlable by Googlebot and indexable for AI Overview purposes. The metered access must be configured to allow Googlebot access without the meter applying – bots should receive full content on every crawl visit, not just the first. Publishers using metered paywalls should verify through Search Console URL Inspection that Googlebot receives full page content.
Content behind a hard registration wall – requiring email address before any access – depends on whether Googlebot can bypass the registration requirement. Typically it cannot without explicit publisher configuration. Registration walls that trigger before any content is served are functionally equivalent to hard paywalls for AI Overview eligibility purposes.
The Strategic Question of Whether to Open Content to Increase AI Overview Citations
The trade-off for subscription publishers is documented clearly. Opening content entirely maximizes citation eligibility but accelerates zero-click loss – users get their answer from the AI Overview without visiting the page or encountering a subscription prompt. The citation delivers brand exposure without traffic or subscription conversion opportunity.
Pew Research in March 2025 found users clicked through to cited sources in only 1% of all visits when an AI Overview was present. Getting cited produces brand awareness but minimal revenue impact for paywalled publishers. AI-referred traffic converts at 4.4 times the rate of traditional organic search – but 1% of queries where an AI Overview is present generates clicks means the absolute volume of AI-referred subscription conversions is small.
Publishers with subscription-dependent revenue models cited AI Overview citation as a net negative even when they held the citation. The Forbes case – 44,131 AI Overview mentions yet 50% traffic loss – illustrates the scale of the trade-off. Citation volume does not protect revenue when the citation format eliminates the click that would have generated subscription exposure.
The strategic question is audience quality, not citation volume. AI-referred traffic that does arrive converts at 4.4 times the organic baseline – it is the highest-intent traffic entering the funnel. For publishers where each conversion has high lifetime value, even low click volume from AI citations may justify maintaining open access on high-value informational content.
Hybrid Content Models That Balance Revenue and AI Overview Visibility
The hybrid content architecture that balances visibility and revenue: keep high-value informational content open for AI Overview citation and organic traffic; put investigative reporting, premium analysis, and time-sensitive news behind the paywall.
This preserves AI Overview presence for the content types most likely to earn citations – informational, educational, definitional – while protecting monetizable content that has direct subscription conversion value. Investigative reporting and premium analysis are rarely cited in AI Overviews because their value is in depth and exclusivity that cannot be extracted in 40 to 60 words. Making these open would not increase AI Overview citation rate meaningfully.
Practical implementation: audit your content by type and paywall status. Identify which pages receive AI Overview citations and whether those pages are currently open or gated. If cited pages are already open, the hybrid model is working. If cited pages are gated via Flexible Sampling, evaluate whether the subscription friction that click-through users encounter reduces conversion relative to fully open access.
For non-subscription publishers, open access is the correct default. All optimization effort applied to a paywalled page on a non-subscription site – where no revenue comes from the subscription prompt – is wasted. Open access removes the only structural barrier to AI Overview citation that is not related to content quality.
Boundary condition: Google’s December 10, 2025 Search Central documentation update on paywall handling is the most current official guidance. The Flexible Sampling access model requires ongoing publisher-Google relationship management for major publishers. Smaller publishers without established Flexible Sampling agreements should treat open access as the default AI Overview eligibility path. Paywall policy changes by Google could shift the citation balance between open and gated content.