Do AI Content Licensing Deals Get You Cited?

Two things were published eleven days apart this month, and read together they answer a question clients have been asking us since March.

The first says AI content licensing deals earn publishers 48% more ChatGPT citations. The second is that Reddit, which holds one of the most expensive licensing deals in the category, lost between 86% and 95% of its ChatGPT citation share in six days.

Both are true. Neither is the headline anyone printed.

Do AI content licensing deals get you cited?

On ChatGPT, yes, on average. Licensed news pages earned 10.2 citations each against 6.9 for unlicensed pages, a 48% premium across 129.3 million citations. On Google AI Overviews, Google-licensed publishers were cited slightly less than comparable unlicensed ones. On Perplexity, licensing made no measurable difference at all.

So the useful version of the answer is that licensing is a one-engine effect, not a property of AI search. And it is an average, which is a different thing from a guarantee.

A licensing premium is a statistical tendency across a corpus. It is not a floor under any single domain.

What the 129.3-million-citation study actually measured

Press Ranger and OtterlyAI published the study on August 20 and 21. It is the only measurement of this question that exists, so it is worth being precise about what it covers and what it does not.

The sample is the widest citation panel published to date

The study captured 129.3 million citations across more than 20 million cited URLs in June 2026, on seven platforms: ChatGPT, Google AI Overviews, Google AI Mode, Perplexity, Microsoft Copilot, Gemini and Claude.

Those citations were matched against every confirmed AI-publisher licensing agreement. The release states that 91 such deals have been signed since 2023, one of them reported at up to $50 million a year.

The premium is real and it is OpenAI-specific

Across all seven platforms combined, licensed pages averaged 10.7 citations against 7.3 for unlicensed, a 46% lift. On ChatGPT alone the gap widens to 48%.

The sharper cut: publishers that signed with OpenAI and nobody else earned 112% more citations per page on ChatGPT. OpenAI is the only licensor in the study whose deals produced a clear advantage on its own platform.

Two engines show no premium, and one shows a slight penalty

This is the part that got left out of every write-up. Google-licensed publishers were cited on Google AI Overviews at a slightly lower rate than comparable unlicensed publishers. Perplexity's licensed publishers landed at parity on Perplexity.

If licensing worked the way the headline implies, all three would point the same way. They do not.

The study is a joint release by two commercial partners

Press Ranger and OtterlyAI have been partners since November 13, 2025, in a partnership announced for the express purpose of improving AI search visibility through PR.

The finding supports the thesis both companies sell. That does not make the numbers wrong, and we are using them. It does mean this is not independent replication, and the study publishes no confidence intervals, no control for publisher size or domain authority, and no stated limitations. Cite it with the partnership attached.

5 reasons a licensing deal is not AI visibility insurance

The premium exists. What follows is why it should not change what a brand does on Monday.

Reason #1: The most-cited domains in the same dataset hold no deals

The leaderboard in the study's own data includes NerdWallet, Healthline and Bankrate. None of them has a licensing agreement with anyone.

A contract shifts the odds on a page. It does not gate the top of the list. The sites winning most of the citations got there without one.

Reason #2: Format is doing more work than the contract is

Best-of lists, buying guides and product reviews account for 46.9% of licensed publishers' citations. That is nearly half the benefit concentrated in three commercial content formats.

Licensed publishers are not being cited because they are licensed and then also happening to publish buying guides. The buying guides are the asset. We found the same pattern from the other direction in do blogs or product pages get cited by AI.

AI does not cite contracts. It cites the format that answers the question.

Reason #3: The most expensively licensed source in the category got wiped anyway

Reddit holds an OpenAI content licensing deal reported at roughly $70 million a year, a figure neither party has confirmed. It is derived from Reddit's own disclosure that AI licensing is about 10% of revenue, minus Google's confirmed $60 million.

Between August 8 and August 14, 2026, Reddit's share of ChatGPT Search citations fell 86.4% by PromptWatch's panel and 95.3% by Qwairy's. No announcement, no explanation, no recourse. We worked through both panels and what they disagree about in the Reddit collapse post.

June's data says the licensing premium was real. August says it protected nothing.

Reason #4: You almost certainly cannot buy one

Roughly 20 news publishers hold OpenAI agreements, covering more than 160 outlets. The confirmed values run from $16 million a year at the floor to $250 million over five years at the ceiling.

These are contracts between model labs and news organisations with archives worth licensing. A B2B software company has no archive to sell and no counterparty interested in buying it. For almost every brand reading this, the licensing lever does not exist as a lever.

Reason #5: The premium was measured once, in one month, on one snapshot

June 2026 is a single capture. The Reddit event demonstrates that a source's standing in one engine can move by an order of magnitude inside a week, which means a one-month snapshot of anything in AI search has a short shelf life.

We measure the normal, non-dramatic version of this drift constantly, and it is covered in why your AI visibility changes weekly. A number captured in June is a description of June.

What a licensing deal buys:

• •A measurable average citation lift on ChatGPT
• •Content reaching the model through a contract rather than a crawl
• •The ability to block crawlers and still be cited by that platform
• •Nothing on Perplexity, and slightly less than nothing on Google AI Overviews

What a licensing deal does not buy:

• •Protection against a retrieval change at the licensor
• •A position on the most-cited list, which is held by unlicensed sites
• •Any effect on the six platforms that are not ChatGPT
• •Notice, explanation or appeal when the share moves

What our own corpus says about who gets cited

We ran the CITE Index for 63 days, from May 19 to July 21, 2026, putting 500 buyer prompts through ChatGPT, Gemini and Google AI Mode every night. It produced 90,132 AI answers across 10 consumer categories before we concluded it. None of the brands in it held a licensing deal.

Four of the twelve most-cited domains in that corpus were brand-owned sites. The other eight were publishers, communities and marketplaces. Reddit was the single most-cited source, drawing 14,698 citations and appearing in 13.6% of all answers.

The engines also cite at very different base rates before anyone signs anything. Google AI Mode cited a source in 97.4% of our answers, ChatGPT in 92.5%, Gemini in 79.1%. A brand tracked on Gemini starts with roughly one answer in five that cites nobody.

And the standings barely move. The average category leader appeared in 78.2% of its category's answers, the leader flipped on only 18.7% of day pairs, and in four of ten categories it never changed once. The full corpus is in the final report.

That last figure is the one that matters here. Incumbency in AI answers is held firmly, and it was held by sites that bought nothing.

The question is not what you can buy your way into. It is what the engine is already reaching for.

5 steps to earn the licensing premium without a licensing deal

The diagnostic half is done. This is the part that transfers to a brand that will never sign a content deal with a model lab.

Step 1: Find out which licensed and unlicensed publishers answer for your category

Run your twenty highest-intent buyer prompts through ChatGPT Search and list every domain cited. Mark which of them are news publishers with known agreements and which are not.

Most B2B categories come back dominated by trade publications, review platforms and community threads, none of which are licensed. That list is your actual target set.

Step 2: Get placed inside the formats that earn 46.9% of the citations

Best-of lists, buying guides and product reviews carry nearly half the benefit in the study. Those formats exist in every B2B category, on trade sites and review platforms, and getting into them is an earned-media job rather than a legal one.

This is the closest available substitute for a licensing deal and it costs a fraction of one. We ranked the surfaces worth pursuing in off-page citation placement with no domain authority.

Step 3: Publish the buying guide the licensed publishers would have written

If the format is what gets cited, the format is also what you can build. Comparison tables, evaluation criteria, priced tiers, honest limitations.

Licensed publishers do not have a monopoly on writing a good buying guide. They have a distribution advantage, which internal linking and earned placement can partly close.

Step 4: Split your citation reporting by engine, permanently

Licensing moved ChatGPT by 48%, Google AI Overviews slightly negative, Perplexity zero. A blended cross-engine visibility score would have averaged those three into a meaningless single number.

Any effect worth acting on in AI search is engine-specific until proven otherwise. Report one line per engine, which is the case we make in how to measure GEO and AI visibility.

Step 5: Set the baseline now, because the next revaluation will not be announced

Reddit's share moved by an order of magnitude in six days and the only reason anybody knows is that two vendors happened to be logging it. The trade press was twelve days behind the data.

Continuous measurement is the only thing that turns an event like that into a decision instead of a surprise, and it is the reason a managed GEO agency runs the panel weekly rather than shipping a quarterly PDF.

What to do, by what you actually are

Your situation	What licensing means for you	Where the effort goes
B2B SaaS with no publishable archive	Not available at any price	Earned placement in the buying guides and review platforms already cited for your category
News publisher weighing an OpenAI deal	Real lift on ChatGPT, nothing elsewhere	Negotiate it, then keep measuring, because the deal is not a floor
Publisher already licensed to Google or Perplexity	Parity or slightly worse on that platform	Treat the deal as revenue, not as a visibility strategy
Brand blocking AI crawlers today	Blocking removes the only route you have	Open crawl access first, since you have no contract carrying content in
Anyone quoting the 48% in a business case	One month, one snapshot, two partners	Attach the date and the partnership disclosure before it reaches a slide

FAQ

Do AI content licensing deals increase citations?

On ChatGPT, yes, by 48% on average. Licensed news pages earned 10.2 citations each against 6.9 for unlicensed pages across 129.3 million citations captured in June 2026. The effect does not generalise: Google-licensed publishers were cited slightly less on Google AI Overviews than comparable unlicensed publishers, and Perplexity's licensed publishers landed at parity. Licensing is a one-engine effect measured in one month.

How many AI content licensing deals are there?

There were 12 publicly announced deals in 2023 and 91 confirmed agreements between AI companies and news publishers by the end of 2025, with roughly 127 projected by mid-2026. OpenAI holds about 20 publisher agreements covering more than 160 outlets. Only three of those have public values: $16 million a year for Dotdash Meredith, more than $250 million over five years for News Corp, and an undisclosed sum in the tens of millions of euros for Axel Springer.

Can a B2B company get an OpenAI licensing deal?

Effectively no. These agreements are struck between model labs and news organisations that hold large archives worth licensing for training and real-time grounding. A software company has no comparable archive to sell and no counterparty seeking to buy one. The transferable part is the format finding, since best-of lists, buying guides and product reviews account for 46.9% of licensed publishers' citations, and those formats are open to anyone.

Does a licensing deal work if you block AI crawlers?

For the licensed platform, yes. Content covered by a deal reaches the model through the contract rather than the crawl, so a licensed publisher can block crawlers and still be cited by that platform. For every other engine, blocking removes the only route in. Any brand without a contract that blocks crawlers is choosing invisibility, which is the failure mode we cover in where AI citations come from.

Did Reddit's OpenAI deal protect its citations?

No. Reddit holds an OpenAI licensing deal reported at roughly $70 million a year, unconfirmed by either party, and its share of ChatGPT Search citations still fell between 86.4% and 95.3% between August 8 and 14, 2026. Two independent panels measured it, both label the finding provisional, and neither Reddit nor OpenAI has commented. The contract bought an aggregate premium in June and no protection in August.

The bottom line

The study is good and its headline is narrow. Licensing buys a 48% citation premium on ChatGPT, roughly nothing on Perplexity, and slightly less than nothing on Google AI Overviews. It was measured once, in June, by two companies that sell the conclusion.

Then the most expensively licensed source in the entire category lost most of its ChatGPT citations in six days, and nobody was told why.

If you are a publisher, sign the deal and keep measuring. It is revenue with a citation bonus attached to one platform, and it is not a floor.

If you are anyone else, the transferable finding is not about contracts at all. It is that 46.9% of the benefit landed in three commercial content formats, and the most-cited domains in the study were the ones that never signed anything. Those two facts describe a route that is open to you, and it starts with knowing which sources are answering for your category right now. An AI visibility audit will tell you that in a fortnight, which is faster than any contract negotiation has ever gone.