Google AI Mode 查询走向上层漏斗，付费搜索投放需调整评估方式

Google’s head of search data science told advertisers on Oct. 28 that people now come to Google earlier in their buying journeys, asking long, detailed questions that start a long way from a purchase. That same afternoon, Alphabet’s chief business officer told Wall Street that the AI Mode queries driving that shift make Google less money than traditional searches do.

Put those two statements side by side, and you have the planning problem facing every paid search manager heading into 2027. The queries are arriving at the top of the funnel, while most Google Ads accounts are still judged at the bottom.

This column pulls the advertiser lessons out of that conversation, checks its numbers against what Google has published elsewhere, and turns it into four changes you can make in your accounts before Black Friday.

An Ads Podcast That Went Organic On Purpose

On Google’s Ads Decoded podcast, Oct. 28, host Ginny Marvin opens by saying the show is “getting out of the advertising world and into the organic search world” for the day.

Her guest is Shivani Mohan, Google Vice President of Data Science and User Experience Research (UXR) for Search. Mohan’s UXR teams interview people one-on-one to learn why they phrase questions the way they do, and her data scientists study logs of what billions of users actually do. She started her career designing bicycles, where she says, “you get one shot to really nail the user understanding” (3:04) because a finished product can’t be iterated.

Google can iterate on Search every day. Advertisers, on the other hand, get roughly one shot per budget cycle to convince finance that a channel is working.

What Mohan’s Team Is Seeing

Mohan calls this “an expansionary moment” for search. For most of Google’s history, she explains, people had to translate a curiosity into a keyword. That friction is fading, so “people are asking much longer, much more detailed questions.” More important for keyword lists, she says, “They’re asking net new types of queries that they would have never thought to ask before.”

Marvin asked the question many PPC managers would ask. Are people coming back with follow-ups because the first answer failed?

Mohan said no, and gave a detail I hadn’t heard Google explain before. “In our data science practice, we penalize queries, which we call man refs, manual refinements,” because a user having to add words means Google missed the original intent. By her account, the growth in follow-ups reflects new curiosity rather than bad answers.

For advertisers, that follow-up behavior is no longer just a user experience story. Google says it is launching multi-turn ads in AI Mode in the U.S., so ads can appear in follow-up queries, according to Shashi Thakur, VP and GM of Google Search Ads.

“What we’re seeing more and more of with AI is that people’s queries are kind of moving up the funnel,” Mohan said, with more brainstorming questions and more questions about how to think about a category altogether.

She also described what shoppers do when they get close to a decision. When people compare two products, “they really don’t want to just see the summary.” They want reviews, podcasts, and videos from people in their own circumstances.

Check These Numbers Before They Go In Your Next Deck

Mohan says query length has increased three times, and Marvin repeats it in her wrap-up as queries in conversational search “running three times longer on average.” Google’s published version is narrower. Its AI Mode Report (U.S., May 2026), as cited by Thakur, says queries in AI Mode are on average three times longer than traditional searches. That is an AI Mode comparison, not a claim about all Google searches.

Mohan also says, “one in every six searches is multimodal.” Google’s May report put it at more than one in six AI Mode searches in the U.S. And on the Q3 call in October, CEO Sundar Pichai gave a newer, global figure, saying “one in five AI Mode searches globally use voice, images or documents.”

To her credit, Mohan explained how Google vets the numbers it releases. Her team checks that an insight is repeatable, and “we don’t want to be able to put out any number that is at a peak point or is something that is liable to change.”

I take her at her word on that rigor. But the durable number advertisers need most is one Google hasn’t published, which is how ads in AI Mode actually perform. Alphabet’s 10-Q doesn’t break out paid clicks or cost-per-click for AI Mode. The only performance figure in Thakur’s post is that click-through rates are “already similar between sponsored and organic product list items within AI Mode,” based on Google internal data from Aug. 20 to Sept. 3, 2026. That’s two weeks of data, measured by Google, comparing ads to organic listings rather than to traditional Shopping ads. Treat it as Google’s claim until your own accounts confirm it.

What The Earnings Call Added

The podcast and the earnings release went out the same day, and the release is where the up-funnel story gets a price tag.

Alphabet reported Google Search & other revenue of $63.35 billion in Q3 2026, up 12% year-over-year, after growth of 19% in Q1 and 17% in Q2. Sequentially it was nearly flat, up from $63.27 billion in Q2. The 10-Q, filed Oct. 29, shows paid clicks on Google Search & other up 8% year-over-year, down from 13% in each of the first two quarters, while cost-per-click rose 4%.

Meanwhile, Pichai said query growth was the highest it has been in five years, including commercial queries.

When JPMorgan’s Doug Anmuth asked why queries were growing faster than Search revenue, Google’s SVP and Chief Business Officer Philipp Schindler pointed first to currency, which he called the largest driver of the slowdown versus Q2. Then he added, “The strong growth in AI Mode queries had an impact because of the lower monetization we were seeing in AI Mode.”

Schindler’s optimistic case lines up neatly with Mohan’s research. “On AI Mode, many of these queries are upper-funnel queries, allowing us to capture new high-value commercial intent,” he said, adding that “the majority of AI Mode queries are incremental, creating more monetization opportunities over time.” Google is now scaling ads in AI Overviews and AI Mode to more users and queries and expanding AI Mode ads to international markets.

Pichai offered his own historical frame on the call, comparing this moment to the shift to mobile, when Google focused first on the user experience and then on monetization. Anyone who managed paid search in that era remembers how it felt. In Q3 2016, Alphabet reported paid clicks up 33% and cost-per-click down 11%, according to Reuters, as searches moved to smaller screens with cheaper ads.

Here is where I think the analogy breaks. In the mobile transition, the monetization gap showed up as cheaper clicks. This time, cost-per-click is still rising, and the gap shows up as queries growing faster than paid clicks. Advertisers aren’t getting a discount for the move up the funnel. They’re getting new top-of-funnel inventory, priced by the same auction, and the burden of proving it pays off.

Marvin’s Advice, Translated Into Account Settings

Marvin closed the episode with her own takeaways for advertisers, starting with a call to “recalibrate how they’re thinking about search” outside of last-click action. Here is how I’d turn that into work this week.

1. Confirm you’re eligible for the new inventory at all. Marvin reminded listeners that “ads in AI mode and AI overviews are matched to both the query and the content of the AI response.” Google’s help documentation for ads in AI Overviews says that only broad match keywords or keywordless targeting (AI Max for Search, Performance Max, Shopping, and Dynamic Search Ads) can serve ads inside the overview. Exact match can still serve above or below it, and Marvin has clarified that an exact match keyword won’t block the broad match version from triggering inside the overview. If your core non-brand campaigns run exact and phrase only, you’re opting out of the placements Google is scaling. Keep exact match where you need control, and give broad match or AI Max a defined slice of non-brand budget with Smart Bidding and clean conversion data behind it.

2. Fill the six conversational attributes in Merchant Center before Black Friday. Marvin flagged these optional fields, which Google rolled out at Google Marketing Live in May. They are question_and_answer, document_link, related_product, item_group_title, variant_option, and popularity_rank. They don’t affect product approval, and Google recommends adding them through a supplemental data source, so your primary feed stays untouched. Mohan’s point about comparison shoppers wanting real-world detail maps directly to question_and_answer and related_product. Start with your top sellers, since Black Friday falls on Nov. 27 this year.

3. Build a long-query segment in your reporting. Export your search terms report and group terms by word count, for example, one to three words, four to seven, and eight or more. Then compare conversion rate, cost per acquisition or ROAS, and how often each group appears in conversion paths. If the long queries convert poorly on last click but show up as earlier touches, you have the upper-funnel evidence finance will ask for. If they don’t show up anywhere, you’ve found where broad match or AI Max is spending on discovery you can’t yet justify, and you can tighten with negatives and brand controls. Either way, don’t let a blended CPA hide the difference.

4. Fix the page the ad sends people to. Marvin said, “Your ad should position your business as the natural next step in that user journey” and urged advertisers to “focus on reducing post-click friction.” A searcher who typed a 25-word question about a specific situation won’t convert on a generic category page. If you use AI Max’s final URL expansion, Google chooses the landing page from your site, so thin pages get picked for nuanced queries. Audit your highest-spend landing pages for the specifics a follow-up question would ask about, such as comparisons, sizing, pricing, and use cases. As Marvin noted, much of this sits outside the PPC team, so bring in your SEO, content, and CRO counterparts.

The Bill For Discovery

Mohan’s bicycle-design lesson applies to advertisers more than she may have intended. Google has the logs, the UXR interviews, and the ability to ship changes daily. Advertisers have the auction bill and a quarterly budget review.

My view is that Google is right about where search behavior is heading and wrong if it expects advertisers to fund that shift on faith. Until Google reports AI Mode ad performance with the same durability standard Mohan described for user statistics, I’d treat AI Mode and AI Overview reach as a test line with its own budget and its own success metric, rather than something that quietly drags down blended CPA.

Google checks its numbers for durability before it publishes them. Your CFO will expect you to do the same.

More Resources:

Featured Image: Elenyska/Shutterstock