Google tells advertisers AI search queries now run three times longer.
A Google search research chief told advertisers on the company's Ads Decoded podcast that queries now run three times longer and one in six searches is multimodal. Google also said ads in its AI search answers are matched to the whole AI response.
The 25-minute episode went up on the Google Ads YouTube channel on 28 October. Shivani Mohan, Google's Search vice president for data science and user experience research, told host Ginny Marvin, the Ads Product Liaison, that people are asking longer, more detailed questions, including new kinds of questions they would not have typed before, as well as questions about images and video and conversational follow-ups. She said Google's data science team counts manual refinements against a result, so it reads most follow-ups as fresh curiosity rather than failed answers. Marvin told advertisers to think beyond last click and pointed retailers to six optional "conversational attributes" in Merchant Center.
The figures are not new. PPC Land, which covered the episode on 1 November, notes that Google's AI Mode report of 19 May, also published by Mohan, gave the same numbers with narrower limits: the threefold length applied to AI Mode queries against traditional searches, and the one-in-six multimodal share applied to AI Mode searches in the United States. On the podcast, neither speaker mentioned AI Mode or the US when giving them.
Plenty is still missing. The episode gave no baseline, market or time period for either figure, named none of the six Merchant Center fields and offered no click-through, conversion or cost data for ads in AI Mode or AI Overviews. Nor did it say how much traffic AI answers send to websites, a figure outside studies suggest is falling.
For advertisers in Saudi Arabia and the Gulf, these are US AI Mode numbers, not a picture of local search behaviour. The direction is still useful: longer, conversational queries matched against whole answers favour rich product feeds and detailed landing pages in both Arabic and English over tight keyword lists. Test that shift against your own results before moving budget, because Google has not published performance benchmarks for these placements.
Why it matters
Google is using behaviour data to steer advertisers towards automated matching and richer product feeds, but the headline numbers come from US AI Mode and carry no performance proof.
Related trends.
X opens lead gen ads to every advertiser, including self-serve
X made its in-feed lead generation ads available to all advertisers worldwide on 30 October, including small businesses and those who buy through Ads Manager themselves. People tap the ad, fill in a short form without leaving X, and the advertiser collects the details.
Why it mattersLead forms are a staple buy on Meta and LinkedIn; X now offers the same route to small and self-serve advertisers, though its only proof point is a single, inconsistent case study.
Reddit ad revenue jumps 54% as it moves to logged-in browsing
Reddit's advertising revenue rose 54% to $847m in the third quarter, but the company warned that falling search referrals could cut its daily users by a low-double-digit percentage this quarter. In response, it plans to make logging in the normal way to use Reddit.
Why it mattersAI answers are draining search clicks from Reddit, so advertisers should expect fewer but better-identified users and higher prices per user.
Meta says ranking upgrades lifted Facebook ad conversions by 15.7%
Meta has published new figures on how its AI ad systems are performing, saying changes to ad ranking drove 8.3% more clicks and 15.7% more conversions on Facebook. The numbers came in an announcement on 28 October, reported by Social Media Today.
Why it mattersMeta's delivery models now do more of the targeting work, so feeding them broad audiences, varied creative and clean conversion data matters more than manual audience tricks.
