More than one-quarter of search queries that have eight or more words in the query now trigger AI Overviews. Many of them now have more than one, identifying them as multi-intent queries.
“If you think about asking the AI search engine questions -- eight-plus words can get very conversational -- and those long queries are the ones AI can help with the most,” said Jim Yu, chief executive officer of BrightEdge.
BrightEdge data shows AI has begun to drive more ad impressions in search engines. He said it makes the media buy more valuable as the person searching dives deeper into the topic and intent.
Someone asking the engine for general questions is valuable, but once they start querying portability, cost and RAM -- getting in front of that buyer is worth much more.
"Keyword-driven intent made the click became much more valuable," he said. "I'm not sure how Google will do it, but you will be able to get better targeted media buys against shoppers further down the AI conversation."
advertisement
advertisement
AI Overviews has evolved a lot since it initially launched in May 2024.
First it handled a lot of simple questions and definitions. Then it evolved into having more than an opinion. Now there are multi-intent queries.
Multi-intent queries include “which CRM is the best for a mid-size business” and “how much time will it take me to implement.” One query serves multiple questions with intent. Another question might be “what are the local vegan restaurants that can serve 10 people at one table.”
This is because Google has precise datasets, including reviews, and known most attributes in local businesses.
“Multi-intent, precise complex questions where help is needed, Google is tuning their AI with all their information to answer questions,” Yu said. “It’s one of the biggest changes we are seeing.”
Local intent is something else Yu has noticed Google has been “playing” with in AI Overviews.
He said Google had increased local citations within AI Overviews significantly up until the end of 2024, and then in January 2025 began to decrease the amount.
“They had ramped it up in the 40s in terms of percentage, which is a lot, and then scaled it back to about 24% or 25%,” Yu said. “When AI Overviews first came out there was no local citations within AI Overviews.”