Do AI Licensing Deals Increase Citations? What the OtterlyAI and Press Ranger Study Shows
OtterlyAI and Press Ranger found a sizeable citation premium for publishers with OpenAI licensing deals, particularly in ChatGPT. The research is an important signal, but its authors caution that it cannot prove licensing caused the advantage.
A new study from OtterlyAI and Press Ranger finds that publishers with OpenAI content-licensing deals received more citations in AI search results than unlicensed publishers in its dataset. The gap was particularly pronounced on ChatGPT. But the result should not be read as proof that signing a licensing agreement directly produces more visibility.
The distinction matters for publishers, PR teams, and search professionals trying to understand how generative search selects sources. Licensing may be part of the picture, especially on the platform operated by the company that made the deal. Yet publisher scale, content formats, cross-platform agreements, and the domains included in the analysis can all affect citation patterns.
What the study found
The August 2026 research examined a June 2026 snapshot of 129.3 million AI-citation events across more than 20 million URLs. OtterlyAI combined those observations with Press Ranger's record of 91 confirmed AI-content licensing deals associated with 314 publisher domains.
The analysis covered seven platforms: ChatGPT, Google AI Overviews, Google AI Mode, Perplexity, Microsoft Copilot, Gemini, and Claude. Its strongest reported relationship concerned OpenAI-licensed publishers on ChatGPT.
According to the OtterlyAI and Press Ranger whitepaper, pages from OpenAI-licensed publishers averaged 10.2 citations per cited page on ChatGPT, compared with 6.9 for unlicensed pages. That is a 48% per-page premium in the study's dataset.
Across all seven platforms, licensed pages averaged 10.7 citations per cited page versus 7.3 for unlicensed pages, a 46% difference. The study also found that 57.9% of all AI citations received by OpenAI-licensed publishers appeared on ChatGPT. Licensed publishers as a broader group had a ChatGPT-dominant profile, with 55.0% of their citations attributed to ChatGPT.
| Measure in the study dataset | Licensed pages | Unlicensed pages | Reported difference |
|---|---|---|---|
| ChatGPT citations per cited page for OpenAI-licensed publishers | 10.2 | 6.9 | 48% premium |
| Citations per cited page across seven AI platforms | 10.7 | 7.3 | 46% premium |
These figures describe citation events per page that was cited. They do not establish an overall share of every AI search, user clicks, referral traffic, conversions, or revenue.
Why the data does not establish causation
The authors explicitly frame the findings as an association rather than proof that licensing causes citation gains. That is the central limitation when interpreting the headline numbers.
A licensing deal may be connected with stronger citations because the agreement changes a platform's access to, use of, or preference for a publisher's material. It is also plausible that publishers able to secure licensing deals were already more visible, larger, more authoritative, or better represented in the query categories AI systems commonly answer.
Several confounders identified in the research deserve attention:
- Publisher size and portfolio scale: Large media groups may own numerous high-performing domains and publish at a volume that naturally creates more opportunities to be cited.
- Cross-platform licensing: A publisher can hold agreements with more than one AI company. That makes it difficult to isolate the effect of a single deal or platform.
- Domain selection: The dataset maps deals to 314 domains, and 90.8% of cited pages on licensed domains were on domains explicitly named in deal announcements. Which domains are included can affect comparisons.
- Content mix: Licensed publishers in the dataset were heavily represented in evergreen and commercial content. Best-of and top-list articles, buying guides, money advice, and product reviews accounted for 58.8% of their citations.
In other words, the study identifies a meaningful pattern worth monitoring, not a guaranteed mechanism. A map of deals and citations can show where the two overlap, but it cannot by itself show why the overlap occurred.
The platform picture is not uniform
The results do not support a blanket conclusion that any AI licensing agreement improves citations on every AI platform. The strongest relationship reported was between OpenAI licensing and ChatGPT citations.
Google-licensed publishers showed only a slight advantage, or a negative result in some measures, on Google AI Overviews. Perplexity showed parity between the relevant groups. Microsoft Copilot appeared to favor licensed publishers as a class, although the study attributes that pattern in part to portfolio scale and cross-license dynamics.
That variation is important. AI discovery is not one homogeneous channel. Each platform can use different retrieval systems, source-selection approaches, product experiences, and relationships with publishers. A publisher considering licensing should assess the strategic value of the agreement itself, rather than assuming it creates a universal citation advantage.
What publishers and SEO teams can do now
The study's scope is most relevant to news publishers and evergreen content, not to time-sensitive news coverage alone. For teams working on AI visibility, it provides several practical areas to investigate.
First, separate reporting by platform where possible. A rise in ChatGPT citations does not necessarily predict a rise in Google AI Overviews or Perplexity. Tracking brand mentions, source citations, and earned media by platform can reveal whether visibility is concentrated in one environment.
Second, review the formats that AI systems are already citing within a site's competitive set. The study's licensed-publisher dataset had substantial representation from comparison-style and commercially useful formats, including product reviews, buying guides, financial advice, and ranked lists. That does not mean these formats automatically earn citations, but it suggests that useful, structured evergreen material deserves attention alongside breaking coverage.
Third, do not treat a licensing deal as a substitute for content quality or technical accessibility. Unlicensed sources continue to appear in AI-generated answers. Clear sourcing, distinctive expertise, well-organized pages, and content that directly serves the underlying query remain important whether or not a publisher has commercial agreements with AI companies.
For a practical baseline, create a GEO Search Leads account to assess your website's GEO readiness and identify optimization opportunities around the content and entities most relevant to AI-driven discovery. That can help teams distinguish a broad visibility issue from a platform-specific pattern before changing editorial, PR, or technical priorities.
Frequently Asked Questions
Does this study prove that AI licensing deals cause more citations?
No. The study reports an association between licensing status and citation frequency. Its authors state that publisher size, cross-platform deals, domain selection, and content mix may contribute to the observed differences.
Which platform showed the clearest licensing-related citation pattern?
ChatGPT showed the clearest reported pattern. Pages from OpenAI-licensed publishers averaged 10.2 citations per cited page, compared with 6.9 for unlicensed pages in the study dataset.
Did Google-licensed publishers receive more citations in Google AI Overviews?
Not consistently. The study found only a slight edge for Google-licensed publishers in some measures and negative results in others for Google AI Overviews.
What types of content were common among citations to licensed publishers?
Evergreen and commercial formats were prominent. Best-of and top-list content, buying guides, money advice, and product reviews represented 58.8% of licensed-publisher citations in the study.
Should an unlicensed publisher assume it cannot appear in AI answers?
No. The findings do not say that licensing is required for AI citations. Unlicensed sources remain part of AI search results, and the research does not establish that licensing alone determines source selection.
Conclusion
The OtterlyAI and Press Ranger study offers a strong reason to examine licensing status as one possible factor in AI citation visibility, especially for ChatGPT. Its evidence is correlational, however, and platform results vary. The most useful response is disciplined measurement: compare visibility by AI platform, understand the content formats earning citations, and avoid treating any single commercial relationship as a guaranteed discovery advantage.