Generative AI mania is transforming how users find information on the internet. And, as referrals from traditional search drop off, some publishers are scrambling to strike licensing deals that will ensure their content gets surfaced – and cited – by AI tools.
Financial news outlet Benzinga isn’t new to the licensing game, said Clint Rhea, manager of institutional partnerships. Licensing its content and its data on financial markets has long been one of Benzinga’s three main revenue streams, alongside live events and digital revenue from advertising and subscriptions.
But until the rise of AI search, Benzinga mostly licensed its content and data to banks and investment firms. Now, Benzinga sees a chance to get its content directly in front of any user who’s looking for it by licensing it to a wider range of generative AI app developers, Rhea said.
With that in mind, Benzinga struck a deal with Dappier, which operates a marketplace of AI tools and licenses data from publishers to train those tools. Dappier ensures that chatbot responses using Benzinga’s data link back to the original source, and it also shares revenue from ads placed in these responses. The business model exemplifies a growing trend of startups helping publishers monetize AI search.
Scaling up content licensing
The deal is a continuation of Benzinga’s commitment to delivering dependable financial information to consumers “wherever they’re at,” Rhea said.
And, he added, there’s no use fighting increased AI adoption “because it’s happening; we’re all using it.” So publishers need to lean into users’ changing habits to stay relevant, he said. “We want to be at the place where people are asking questions, and we want to be the source to deliver that information.”
Through Dappier’s marketplace, a variety of AI tools, from search chatbots to specialized AI agents, can ingest Benzinga’s data. For example, the data could be surfaced by a chatbot responding to a user query about the best up-and-coming stocks to watch. Or it could be used to train a bidding algorithm for an investment bank with real-time updates from Benzinga’s coverage of the stock market. Or it could be used by an ad agency to create an AI agent tasked with finding new ways to target with ads people who work in finance.
Through an RSS feed integration, Dappier ingests content from Benzinga as it’s published. It can also pull non-content data – like historical fluctuations in certain stock prices – through API connections. Then, through retrieval-augmented generation, Dappier provides AI app developers access to Benzinga’s data set for use by their tools.
The arrangement gives Benzinga the support it needs to scale up its content licensing business, Rhea said.
Benzinga fields “dozens of emails a day from people wanting to license our content,” he said. Now, Benzinga has Dappier’s help vetting those deals and monitoring how the AI tools use its data over time.
Dappier also requires that any apps that rely on Benzinga’s content provide proper attribution and link back to the source where appropriate, Rhea said. Plus, Dappier provides Benzinga transparency into who’s licensing its data and for what purposes, he said. Benzinga also has veto power to deny developers access.
And if Benzinga ever wanted to reverse course, Dappier would purge all of its data out of the AI marketplace, said Dan Goikhman, co-founder and CEO of Dappier – including data that had already been used to train AI models.
A new ad channel for publishers
Publishers earn revenue from their Dappier integrations via data usage fees and advertising.
If an AI app developer ingests Benzinga’s data through Dappier’s marketplace, then Benzinga collects a fee each time its data is used in response to a user query.
The increased use of Benzinga’s data in AI solutions is also generating a new stream of ad revenue, which is notable, Rhea said, because “advertising, historically, has not been a large portion” of the revenue mix for Benzinga’s licensing department.
Dappier shares revenue with the publisher from ads placed alongside query responses that pull from the pub’s data. It serves ads via integrations with Google Ads, Magnite, PubMatic, Sovrn, AdMarketplace and Advertising.com.
These ads are all either standard IAB display banners or affiliate-esque native links to recommended products. How they appear changes depending on the interface of the AI tool. Dappier offers placements for ChatGPT-like AI search interfaces, as well as chat widgets embedded in-article or in the site’s right rail.
Speaking of those embedded site widgets, Dappier also provides a service for publishers to embed their own AI tools into their sites – a sort of “ask the publisher” button. Adopting such a tool could be “phase two” of the partnership, Rhea said. Benzinga’s also discussed possibly launching its own standalone AI app, he said.
But, Rhea added, he believes opening up its data to third-party AI developers will let Benzinga make more money in the long run than it would by monetizing its own product.
Another major benefit Benzinga gains from the deal – and from being an early mover in AI licensing – is control over how generative AI solutions are using its data, Rhea said. That means fighting back against the tech’s tendency, to, well, make stuff up.
“When I’m interacting with any of these services, I just want accurate information,” Rhea said. “We want to remove some of those hallucinations that go on and be a trusted resource.”
