Microsoft, Google strategy to test AI search ads irks some brands

The two tech giants are racing to revamp their search engines with artificial intelligence

(Reuters) - Alphabet's (GOOGL.O) Google and Microsoft (MSFT.O) are inserting ads into AI experiments without providing an option to opt out of participation, an approach that has already rankled some brands and risks further pushback from the industry, ad buyers told Reuters.

The two tech giants are racing to revamp their search engines with artificial intelligence that can produce written responses to open-ended queries. That process will upend how advertisers reach consumers through ads on search results, a market that is estimated to grow 10% to $286 billion this year, according to research firm MAGNA.

Microsoft is testing ads in the Bing AI chatbot, which began rolling out to users in February, by relocating some traditional search ads and inserting them into the AI responses, the company said.

In a May interview, Google’s general manager of ads, Jerry Dischler, said the tech company would also use existing search ads to experiment with ad placements within the AI search snapshots, an early-test feature called Search Generative Experience that first became available last month. Google said advertisers are currently not able to opt out of the test.

Both companies said they are in the early stages of testing ads in generative AI features and were actively working with advertisers and soliciting their feedback.

Some advertisers are wary of their marketing budgets being spent on features that are available to a limited number of users, ad buyers said. Advertisers typically also want to have control over where their ads appear online and are cautious about appearing next to inappropriate or unsuitable content.

Microsoft and Google said existing guard rails on their search engines, including lists of blocked keywords to prevent ads from appearing on those queries, would also apply to their AI search features.

The two companies have poured billions into generative AI, making it crucial to derive revenue from the technology. That has included investments in other AI companies, such as Microsoft’s multibillion-dollar bet on OpenAI, creator of the popular ChatGPT, and Google’s $400 million stake in Anthropic, an OpenAI competitor.

'BLACK BOX' Testing new ad placements without gaining consent from brands is a concerning practice for advertisers, said Jason Lee, executive vice president of brand safety at Horizon Media, a media agency that has worked with brands like insurer GEICO and Corona beer. Another ad buyer at a major ad agency also said the practice was not considered standard in the industry.

Multiple large advertisers temporarily pulled their ad spending from Microsoft in response, according to an ad buyer familiar with the matter. Among them, Wells Fargo (WFC.N) continues to keep some of its ad budget off Microsoft, the person added.

Wells Fargo did not respond to a request for comment.

In an interview, Lynne Kjolso, Microsoft’s vice president of global partner and retail media, said the company aimed to make the introduction of the new Bing ad formats as “seamless” as possible for advertisers, without creating more work for them.

Microsoft recently launched hotel ads in the Bing chatbot and is working to introduce ads for other industries, such as real estate, she said.

The concerns from advertisers also play in to a larger tension as tech platforms increasingly offer AI solutions that could produce better results for advertisers but require them to give up some level of control over their ads.

“This is not the first time that Google and Bing have expanded their networks while restricting control for advertisers,” said Samantha Aiken, a paid search supervisor at marketing agency Code3.

For example, many in the industry consider Google’s Performance Max, a tool that uses AI to automatically find the best ad placements across multiple Google products instead of requiring advertisers to set up different ad campaigns, to be an analytical “black box,” she said, as the algorithmic models do not disclose how it determines where to serve ads.

Three ad buyers said they were concerned about Microsoft’s lack of transparency reporting, or reports that can show what search terms triggered a brand’s ads to appear in generative AI experiences, or how the ads performed versus traditional search ads.

While Microsoft representatives have been receptive to the concerns, they have not provided a timeline for when more transparency reporting would be available, two of the ad buyers said.

“Advertisers can’t just go in and pull the report to see how often they're showing (in the Bing chatbot),” said one of the ad buyers, who requested anonymity to discuss private conversations with a business partner.

Kjolso said transparency reports have been a top request from ad agencies, and that Microsoft’s product teams were “working on it with priority.”

“We are absolutely thinking through what additional levers and controls we need to be able to provide to advertisers,” she said, adding that sales teams were actively working with some brands to allay any concerns about where their ads are appearing.

Two media buyers from major ad agencies said there were questions about how the search giants would prevent ads from appearing on AI responses that include "hallucinations" or false information.

Bing’s Web information can act as a "grounding" mechanism for large-language models and actually reduce the risk of hallucinations, Kjolso said.