Why publishers fear traffic, ad declines from Google’s AI-generated search results

As Google expands its new “AI Overview” feature, publishers are wondering to what degree AI-generated search results will negatively impact referral traffic. And while it’s too soon to tell the extent of the damage, some companies are hoping for more transparency from the search giant.

Following AI Overview’s rollout last week, some think the impact could be even worse than feared.

Raptive, which helps sell digital ads for 5,000 independent creators’ websites, initially estimated AI Overviews could cut visits by as much as 25% and cause the industry to lose $2 billion in annual ad revenue. However, Raptive chief innovation officer Marc McCollum now thinks that is “maybe [on the] very low end.” 

According to McCollum, Google’s vision for Overviews seems to be a “far more extreme version” than during the past year of beta tests. However, he said it’s hard to tell without more data from Google — which so far it hasn’t shared. He sees the lack of transparency as a more pressing issue for publishers (for now) than the ongoing quality concerns about AI Overview’s answers, adding that this week’s introduction of ads in AI Overviews is “just another nail in the coffin.”

When Google debuted AI Overview last week, CEO Sundar Pichai said its Gemini model has powered answers for billions of queries through its Search Generative Experience (SGE), which preceded AI Overview. (While Google says AI Overviews leads to more clicks, but so far it hasn’t shared concrete info about what that looks like.)

People are using it to Search in entirely new ways, and asking new types of questions, longer and more complex queries, even searching with photos, and getting back the best the web has to offer,” Pichai said. “We’ve been testing this experience outside of Labs. And we’re encouraged to see not only an increase in Search usage, but also an increase in user satisfaction.”

“I think Google owes publishers that level of transparency because publishers have raised such serious concerns,” McCollum said. “We believe what Google is doing is copyright infringement. We do not think fair use applies.”

To help publishers understand the changes, McCollum said Google could give detailed data about AI Overview as it rolls out: Total users with access to AI Overview, the percentage of queries featuring AI Overviews in various markets, and how the feature influences user behaviors.

Google is still a major source of traffic for news and media companies. About a third of April traffic for the top 100 publishers came from Google, according to data provided by Similarweb.

For the New York Times — where 32.5% of organic traffic in April came from Google — that adds up to around 250 million visits. However, other U.S. publishers are even more reliant on the search engine. Google accounted for 72% of organic traffic to Forbes and 60% of traffic for the websites of USA Today, Business Insider and Newsweek.

A far smaller percentage of publisher traffic is coming from ChatGPT, which which found OpenAI’s chatbot surpassed 100 million daily visits twice last week, according to SimilarWeb. In April, ChatGPT generated 113.5 million visits to other websites, but nearly 80% of that referral traffic was directed to OpenAI’s own website. Of the media companies that did get some traffic, none received more than 1% of outbound traffic. Meanwhile, most of the other 23.4 million visits to non-OpenAI pages were for computing-related websites.

To help publishers track the impact of AI summaries, Semrush — a company that works in SEO, search and analytics — created new tools including one called Position Tracking to monitor any site’s visibility in AI Overview. Another tool, Sensor, monitors how many AI Overviews appear on search results across various industries.

For now, AI Overviews seems to only appear in a small fraction of Google search results. Less than 0.5% of Google search engine results pages show an AI Overview, according to Semrush, with AI Overviews showing up in just 0.1% of news-related search results.

“It will have an impact, but I don’t expect it to be as substantial as many fear,” said Kyle Byers, director of growth marketing at Semrush. “It may even have a positive impact on average since AI Overviews link back to source web pages. An improved experience for Google users could also result in more searches being made, which is another way it could lead to increased traffic for publishers and content creators.”

The changes are all part of the ongoing existential dilemma publishers face with generative AI and what it means to their business model. Some have filed lawsuits against Google, OpenAI and Microsoft alleging copyright infringement. Others have chosen to partner on efforts to train AI models on higher quality content. 

Earlier this week, News Corp and OpenAI announced a new deal valued at $250 million in cash and credits over the next five years. The arrangement will let OpenAI answer questions using content and mast heads from more than a dozen News Corp properties including The Wall Street Journal, Market Watch, and the New York Post. when answering user queries. Other deals announced this week include one with Reddit — which follows a previous AI training deal Reddit made with Google — and other publishers like the FT and Dotdash Meredith.

Pete Brown, research director at the Tow Center for Digital Journalism, recently published a report showing OpenAI has confirmed more than 20 deals with publishers. Overall, he counted a total of 35 confirmed deals and another 17 in reported discussions. 

While Google and OpenAI see publisher deals as a way to improve the quality of AI-generated content, some experts researching AI and media think it’s hard to say how it’ll play out even if large language models get better giving real time answers. There’s also the key question of what happens with publisher allegations of tech companies training their AI models without consent.

“In general, given recent allegations, publishers should be able to know if and to what extent their data has already been used to train AI models,” said Felix Simon, a research fellow in AI and news at Oxford University. “If they reach an agreement, future uses of the data including monetization and product options that might directly compete with publishers’ offerings would certainly be helpful to know.”

Others are keeping tabs on what AI Overview could mean to European publishers. David Buttle, who previously spent a decade at the FT, is currently working on a report to examine that impact. Although it’s too soon to tell, he thinks Google might hold off because of regulatory issues including British Parliament’s May 22 passage of Digital markets, Competition and Consumers Bill. Once AI Overview is deployed, its impact could be uneven depending on hard news and high stakes queries, which might soften the blow for premium publishers. However, it might not be as safe for lifestyle publishers, he said, adding that they should start preparing now to reduce reliance on Google through direct reader relationships and unique established voices.

As more users start and finish their information journeys on Google, Buttle thinks it could lead to fewer ads in aggregate even if a higher percentage are on the search giant’s platforms. That might also lead to less ad inventory overall, especially for publishers, while also increasing the value of Google ads.

“Of course [Google] has the intent and personal data to provide an extremely compelling commercial offering versus a publisher,” Buttle said. “Perhaps even more so with AI Overviews which are likely to result, over time, in a different type of search query.”

Publishers and partners are also worried about other recent changes various aspects of search. Other issues include Google’s crackdown on “site reputation abuse” and other SEO spam. Panayotis Nikolaidis, CEO of couponing company Saving United, said he’s noticed traffic losses in the past week have already been significant, showing millions of lost page views. He understands the importance of improving overall search quality, but thinks changes are harming publishers in the process.

“What people are irritated by is there is just not a lot of guidance from Google right now,” said Nikolaidis, whose company works with major publishers to create curated couponing platforms. “People understand the power dynamic and need to understand what Google wants.”

https://digiday.com/?p=545905

More in Media

The Trade Desk and Yahoo are locked in a trading dispute with a looming deadline

If not resolved, it could result in the industry’s largest indie DSP cutting off access to Yahoo’s video inventory.

Why Getty Images and Picsart are partnering to train a new AI image model

The deal let Picsart train create an AI model using hundred of millions of licensed images to power a new platform for both companies.

How a work platform redesigned BuzzFeed’s former offices after moving in

Work platform Monday.com now occupies the former office space of BuzzFeed.