Longer and more detailed searches have not always been suitable for advertising, particularly when user intent did not clearly match the campaign’s keywords. Google is now trying to open up that space with Gemini models, which interpret the meaning of queries and look for a suitable ad even when the user phrases a question in a more conversational or complex way.
The company discussed this change on July 22, during the presentation of second-quarter 2026 results by parent company Alphabet. Revenue from the Google Search & Other segment rose 17 percent, from $54.19 billion to $63.27 billion. Retail and financial services made the largest contribution to growth, while the company’s management also linked the result to the deeper integration of Gemini into its advertising infrastructure.
Rather than using the model only to generate ads, Google applies it across three parts of the system: assessing ad quality, tools managed by advertisers and ads within new AI-powered search experiences. Philipp Schindler, Senior Vice President and Chief Business Officer at Google, explained that Gemini analyses the nuances of longer queries to identify what the user is actually looking for.
For Shopping ads, this type of matching led to a 20 percent improvement in the delivery of highly relevant ads, according to data presented by Google. The figure refers to the relevance of results, rather than directly to sales growth or conversion rates.
The second part of the system relies on AI Max, a Search campaign automation tool that has moved out of beta and is already used by half a million advertisers. The company says campaigns using AI Max or Performance Max generate an average of 15 percent more conversions or conversion value at a similar return on ad spend.
AAA Auto Club Enterprises was presented as one example, having used AI Max to adapt creative assets to more detailed insurance searches. The company recorded 17 percent more conversions and an 11 percent lower cost per lead. These are the results of an individual advertiser published by Google, rather than an average across all campaigns.
For Google, the broader effect of this type of matching is more significant. During the call with investors, Schindler said that AI Max opens up billions of new searches that previously could not be monetised effectively. Ads are therefore relying less exclusively on exact keyword matching and more on assessing the intent behind a question.
The change also extends to AI Overviews and AI Mode. Users ask more specific and detailed questions in these formats, which the company sees as an opportunity for more precise advertising. Within AI Mode, it is currently testing contextual links shaped by the flow of a conversation, special offers shown while users plan purchases or travel, and the Highlighted Answers format, which places clearly labelled sponsored links inside responses structured as lists.
The published information does not mean that ads have been introduced into the Gemini app itself. In this case, Gemini operates in the background of Google’s existing advertising system, where it aims to better understand queries, expand the number of searches on which ads can be shown and reduce advertisers’ dependence on preselected keywords.
