Users are asking longer questions in AI Mode, Google’s conversational search feature, and using it to explore their decisions earlier, according to the latest episode of the Ads Decoded podcast, released on October 28. Participants include Ginny Marvin, Google Ads Product Liaison, and Shivani Mohan, Search Vice President of Data Science and User Experience Research. Both also offered guidance for advertisers preparing their strategies for 2027.
- Users ask longer questions in AI Mode and explore their decisions earlier.
- Discovery and evaluation now happen within the same conversation.
- Users still seek real-world experience from forums, podcasts, and videos.
- Advertisers should answer questions on price, compatibility, and limits, then measure whether these pages lead to purchases.
A more conversational and exploratory search
In a post published on LinkedIn, Ginny Marvin highlights three key takeaways. The first concerns discovery, which happens earlier in the journey: users are employing conversational search to think and discover options well before making a purchase decision. The second relates to curiosity, which is broadening: according to her, queries in AI Mode are three times longer than those in traditional searches, and one in six searches is now multimodal, meaning it combines text, voice, or images. The post does not specify the reference definitions or measurement period for these figures, however.
The episode notes add that users ask questions three times longer, follow up with subsequent questions, and use voice and images. This allows them to incorporate personal elements and work through complex decisions during the search itself. According to the podcast, discovery and evaluation can happen within the same conversation: a person starts by exploring a problem, then asks about options and their trade-offs as they learn more.
What Google’s numbers say
In its announcement published on October 28, Google reiterates the figure of questions being three times longer and points to a report on AI Mode in the United States published in May 2026. This report states that more than one in six searches in AI Mode, in the US, is multimodal. This scope is narrower than the phrase ‘one in six searches’ used by Ginny Marvin, which should not be extended to all Google searches. Furthermore, the statistic compares the average query length without indicating how differences in users or search tasks affect this comparison.
For advertisers, longer questions can provide more clues about a person’s needs. This figure alone does not indicate stronger purchase intent or better conversion rates. Questions asked when narrowing down options can reveal concerns that a product description or general service page does not address.
Users also seek human opinions
In her LinkedIn post about the episode, Shivani Mohan writes that users appreciate ‘the pared-back clarity’ of AI-generated summaries, but actively seek ‘the rich detail, lived experience, and earned authority’ of the open web. She cites forums, podcasts, and videos as places where these perspectives are sought. She also mentions billions of clicks to the open web, without specifying a period or comparison, so this figure does not indicate whether traffic is increasing or how clicks are distributed among sites.
Her comments describe a demand for direct experience, without establishing that increased production of videos or reviews generates more traffic from Google. For advertisers, a review or demonstration can answer questions about a product’s real-world performance, its limitations, or the reasons for a choice, provided it offers an experience that promotional text does not.
In responses to Ginny Marvin’s post on October 29, two users asked how these behaviors translate into campaign management. The posts and the episode do not provide specific settings for these cases. The podcast does, however, encourage advertisers to examine needs expressed even before the customer knows the product name, to ensure their landing pages address questions about price, compatibility, and limits, and to connect support content to the decision. Finally, it advises them to measure whether visits to comparison or educational pages lead to a purchase, over the normal purchase cycle timeframe, as impressions and click-through rates alone are not sufficient to demonstrate this.
