How LLMs Are Changing Publisher Payments in Search

The longstanding partnership between search engines and content creators is undergoing a dramatic transformation. For years, this relationship operated on a simple exchange: publishers allowed their sites to be indexed, and in return, search engines directed valuable user traffic back to them. This referral traffic was the lifeblood for many, funding operations through advertising and subscriptions. The rapid integration of generative AI and answer engines is fundamentally disrupting this model, keeping users within AI platforms and drastically reducing the clicks that publishers depend on. As a result, new compensation frameworks are emerging, creating a complex and uncertain financial landscape for the future of digital content.

The shift in user behavior is stark and measurable. When AI-generated summaries appear in search results, click-through rates plummet. Data shows only about 8% of users proceed to click any link when an AI overview is present, compared to 15% without it, a drop of nearly half. Furthermore, the incidence of “zero-click” searches, where users get their answer without visiting a source site, has risen significantly. This trend correlates with a steep decline in organic traffic to websites, with billions of visits vanishing. The imbalance is further highlighted by the crawl-to-referral ratio; while traditional search maintains a relatively balanced exchange, some AI companies crawl thousands of pages for every single referral they send back. This collapse in pageviews directly translates to fewer ad impressions, lower subscription sign-ups, and diminished affiliate revenue.

In response, three primary payment models are taking shape. The first is usage-based revenue sharing, where platforms like Perplexity share a portion of subscription revenue with publishers whose content appears in their results. Another initiative, ProRata, offers a 50/50 split through its answer engine. While these models directly tie payment to content usage, the revenue pools are currently small and hinge on converting free users to paid plans.

The second model involves flat-rate licensing deals, often worth millions of dollars. Companies like OpenAI have secured agreements with major publishers such as News Corp, The Atlantic, and The Associated Press. These comprehensive deals typically bundle rights for using archives to train AI models, displaying real-time content with attribution, and providing technology access. This approach, however, risks creating a two-tier system where large, established publishers with extensive archives can negotiate favorable terms, while smaller outlets lack the same leverage.

The third avenue is emerging through legal settlements that set precedents. A landmark settlement between Anthropic and authors, following a complex court ruling on fair use, demonstrated that AI companies are willing to pay substantial sums even as they litigate. This settlement provides a public financial benchmark for other negotiations, though specific terms often remain confidential.

Publishers are reacting in divergent ways. Some are embracing partnerships, viewing them as a necessary new revenue stream and a form of legal protection in the evolving copyright landscape. Executives from companies like Condé Nast and Dotdash Meredith have publicly stated that AI platforms should compensate publishers, framing early deals as strategic positioning. Conversely, other publishers are pursuing litigation. The New York Times’s high-profile lawsuit against OpenAI and Microsoft argues that the companies built a profitable business on the unlicensed use of copyrighted material. Other publishers have refused deals they feel undervalue their work, concerned that accepting low offers now would set a damaging precedent for the future.

This divergence is fostering a new division in online content: the Licensed Web versus the Open Web. The Licensed Web consists of premium, often specialized content available through formal agreements, ensuring attribution and compensation. The Open Web encompasses the vast array of freely crawlable sites, including user-generated content and general marketing material, which may be used by AI with little to no direct payment. This divide creates mismatched incentives, where investment in unique, high-quality journalism might be supported by licensing, while more generic information faces commoditization.

For SEO and content professionals, this necessitates a strategic pivot. The focus is expanding beyond driving clicks to also tracking AI citations and brand mentions within answer engines, even when they don’t generate direct traffic. The decision to allow or block AI crawlers via robots.txt has become a critical business calculation, balancing visibility against potential licensing value or the protection of subscription models. With traditional traffic metrics under pressure, there is a greater emphasis on building direct audience relationships through newsletters, apps, and branded search, while diversifying revenue into subscriptions, events, and consulting.

Uncertainty over sustainable payment models raises profound questions about content investment. Publishers with licensing deals may tailor content for AI training and retrieval, while those without must rely on other strategies. Across the board, revenue declines are forcing difficult choices, leading to staff reductions in newsrooms and potentially reducing the capacity for investigative reporting. The core challenge remains: AI systems are fundamentally dependent on high-quality human-created content for training, yet their outputs often undermine the very traffic that funds the creation of that content.

The path forward is unclear and likely depends on the outcomes of ongoing litigation, potential regulatory action, and market pressures. For now, publishers must navigate immediate decisions about technical access, content strategy, and revenue diversification without a clear map of what will prove sustainable in the long term.

(Source: Search Engine Journal)