At the PPA’s Future of Access event last month, a couple of questions kept returning. Do we have leverage? What is the value and the price of our content? What does a good deal look like? In the emerging market of licensing and monetisation, we need to get more serious about answering these, perhaps even developing the skills required within our teams.

In this issue I look at how value is changing, where the power sits, and how to start building leverage.

A tale of two markets

For a couple of decades in video, music and books, negotiation was cleaner and easier because rights were bounded and usage was trackable against each copyright item.

Price was negotiated and followed the rights bundle (based on dimensions like territory, window, exclusivity and platform type). It was anchored on comparables, the buyer’s expected value, and what the seller could earn elsewhere. It was competitive. An interesting benchmark comes from TV, where some reports suggest that licensing to third-party distributors can bring a major broadcaster or network up to a fifth of its commercial revenue.

Compare this with the findings in the PPA’s AI licensing report published earlier this year: the average multi-year AI deal represents under 2.15% of annual publisher revenue. AI deals often bundle training, display and retrieval rights with no shared price history.

The incentive shift

Journalism has long been advertising-funded, and that translated well to the web. Publishers wanted reach, and Google supplied it in exchange for access to their content.

That is changing. Bots now account for more than half of HTML traffic (Cloudflare, June 2026). This traffic needs to be supported and paid for, but bots are not clicking on the ads. Much of it comes from the AI labs, which generate substantial revenue, but not from advertisers. What they need to operate is talent, compute power and data. Data cannot stay free when advertisers no longer pay for the traffic that carried it. Labs need three scarce things from the data they use:

• Distinct, high-quality human content. In 2024, Epoch AI projected that public high-quality text could be exhausted between 2026 and 2032.
• Fresh content. Models learned to write from past text, but their knowledge stops at the training date. USA Today Co.’s CEO expects real-time content to become more valuable than archives.
• Credibility and grounding. The Open Markets Institute argues publisher content adds value across training, fine-tuning, grounding, current information and the credibility of answers.

Who has the power

Even though the publisher provides the scarce asset for the labs, the power does not exactly (or yet) follow. There are a couple of asymmetries in the market that drive this.

• No reliable tracking. Nobody can easily see who uses content or when, which broke the old deal.
• Irreversibility. Knowledge embedded in a model cannot be retrieved, and content already used erodes leverage.
• Buyers set the price. Google’s AI Contribution Pilot reportedly pays around 100 publishers when their content “significantly contributed” to an AI answer. Payments rest on what Google itself deems valuable, with no explanation of the calculation. The valuer is also the buyer.
• Early terms become norms. Today’s deals are tomorrow’s precedents.

Building leverage in three steps

There is no question that there are challenges, but the emerging market will offer a growing number of options. Publishers need to start gaining an understanding of the value of their content and the leverage they have.

1. Understand the value of your content. There are three methods, each with limits:

• Break-even: traffic loss multiplied by revenue per visit. The estimated loss of revenue to LLMs informs what gap needs to be filled by new revenue streams.
• A content value floor per article or token. This is the cost of production of the article, or a technical representation of the article in the form of tokens, multiplied by the average token price on the LLM market.
• Performance uplift: payment tied to usage. How much revenue would have been made if all the bots crawling the content paid their share (in whichever model the publisher monetises)?

Whichever you use, it is about knowing the economic floor and ceiling. A good deal is somewhere between the two.

2. Understand your leverage. Look back at the three scarce things AI labs need. Scarcity turns into money only if you can credibly withhold. That depends on your pathway (protect, participate or pivot) and on acting collectively, since collective licensing routes raise the standing of smaller publishers.

3. Take control through access. Tools such as pay-per-crawl point to a model built on access rather than copyright: block crawlers by default, then license access through an API or subscription. Keep robots.txt, CDN rules and licence terms saying the same thing. Decide access before price, because few decisions about your IP are reversible.

The takeaway

Protect first, then build leverage, then pursue monetisation models you control. The question for your next meeting: who in your organisation owns the analysis of what your content is worth?