Why some publishers aren’t ready to monetize generative AI chatbots with ads yet
Publishers are testing generative AI technology for all kinds of functions — from onsite search to games. But monetizing those offerings is slow going. Some publishing execs told Digiday they’re not ready to do so by adding advertising to these generative AI products just yet.
Like the rest of the digital publishing world, monetizing these AI experiences falls into two camps: subscriptions or advertising — both for publishers and the AI companies themselves.
OpenAI’s more powerful version of ChatGPT requires a subscription to use and publishers like Ingenio are starting to roll out subscriptions asking users to pay for more interactions with its AI chatbots. On the other hand, BuzzFeed is selling brand sponsorships to its GPT-powered games, and generative AI search engine Perplexity is planning on adding native advertising around the suggested related questions that appear next to the response to users’ questions, such as organic or brand-sponsored questions, according to Adweek.
Testing AI monetization, but not implementing
Outside’s AI-powered chatbot Scout, launched last year, and is only available to paying Outside subscribers. While Scout isn’t being monetized beyond that right now, Katie Cruickshank, director of digital marketing & strategy at Outside, told Digiday adding ads to the experience is “something I believe will happen in later phases.” Cruickshank declined to share how many subscribers Outside has or how many are using Scout.
Josh Jaffe, president of media at Ingenio, said the publisher is taking a different approach and focusing on subscriptions or registrations for its chatbots. Ingenio’s AI chatbot Veda has a registration wall pop up after three prompts. The next iteration of a chatbot feature, called Ether, will ask for a phone number. The service will next ask users to pay for a subscription to purchase messages with the chatbots, Jaffe said. Jaffe declined to share how many people are paying for a subscription to Veda.
Ether is in beta and has been in development since August and will launch in Canada, New Zealand and Australia next month, Jaffe said.
“We’ve thought about advertising and we’re not ruling it out, especially if [we find] people don’t want to subscribe,” Jaffe said.
So why are publishers hesitant to serve ads in these environments? It comes down to low CPMs, and not enough scale, according to Jaffe and Cruickshank.
Advertising in these generative AI chatbot experience won’t be a “sustainable model” long term, according to Jaffe, unless CPMs “go way up.” Building a subscription model for Ingenio’s chatbots also means the publisher will have more control over revenue, the user relationship and distribution, he added.
And Outside says it’s too early for them to advertise on its chatbot, too, given Scout’s scale, Cruickshank said. She declined to share how many people are using the chatbot.
Publishers should determine “how are audiences using these types of platforms” and understanding user behavior before advertising based around them, said Liz Bartges, director of brand engagement at ad agency FerebeeLane, whose clients are exploring ads in AI products, but haven’t bought any.
But BuzzFeed has snagged advertisers for its generative AI chatbot experiences, like Sprite, Serta and Walmart to create sponsored quizzes and shopping chatbot integrations, BuzzFeed president Marcela Martin said in an earnings call in August (she has since left the company). It’s unclear, however, how much money this is actually bringing in. BuzzFeed did not respond to requests for comment for this story.
How publishers are using chatbot data
Cruickshank was optimistic about using generative AI to create curated environments for advertising, using contextually targeted and first-party data to improve ad alignment, she said. Personally identifiable data from Scout isn’t stored, she added.
“We have an understanding of the user and their data for our product analytics — [like] what questions they are asking and what articles they are clicking through to — and we may eventually create targetable cohorts available to advertisers out of that information, but we are not presently,” Cruickshank added.
Jaffe said Ingenio can develop profiles based on demographic data shared by registered users of its AI chatbots. In the next year or so, the goal is to build a model “based on inferences made from conversations” with the chatbots to help build richer profiles of users. But this wouldn’t be used for ad targeting — at least for now.
“That’s not how we’re thinking about it,” he said.
Do ads in generative AI products undermine trust?
There is some debate over whether generative AI products make users more distrustful of these experiences.
Cruickshank doesn’t think so. “Similar to search, ad-supported chatbots are an opportunity to provide contextually relevant and targeted ads based on queries and activity,” she said.
But Bartges isn’t convinced yet. Her hesitation is that the ads she’s seen in generative AI search engines aren’t clearly labeled or discernable as advertising. She likened it to when ads were first added to Instagram’s feeds, which Bartges said felt “disruptive” in the beginning.
“My first thought is going to be, how disruptive is this going to be? Will it start to erode trust with the audiences, the trust that I’ve earned on my platform, if I start to disrupt their experience when [publishers are creating these products] trying to solve a problem and answering [users’] questions in the first place?” she said.
