On the same day in late October, Google said two things that belong in the same sentence. Its search research team told advertisers that people now arrive earlier in the buying journey. Its chief business officer told Wall Street that those earlier queries make Google less money.
Put them together and you get the planning problem for every paid search budget heading into 2027.
Two statements, one afternoon
A detailed column on Search Engine Journal lined up the two conversations. On Google's Ads Decoded podcast, Shivani Mohan, who leads data science and user research for Search, described "an expansionary moment." People ask longer, more detailed questions, including types of queries they would never have typed into a keyword box.
Then the Q3 numbers. Search revenue reached $63.35 billion, up 12% year over year but nearly flat against Q2. Paid clicks grew 8%, down from 13% in each of the first two quarters. Cost per click still rose 4%.
When an analyst asked why queries grew faster than revenue, Philipp Schindler named currency first. Then he added that strong growth in AI Mode queries had an impact "because of the lower monetization." Google's optimistic framing is that these are upper-funnel queries capturing new commercial intent that will monetize over time.
This is not the mobile shift again
Sundar Pichai compared the moment to the move to mobile, when Google built the experience first and the revenue later. I managed accounts during that period. I remember it well, and the comparison breaks in one important place.
In the mobile transition, the monetization gap showed up as cheaper clicks. In Q3 2016, Alphabet reported paid clicks up 33% and cost per click down 11%, according to Reuters. Advertisers got volume at a discount while everyone figured out small screens.
This time, cost per click is still climbing. There is no discount. The gap shows up as queries growing faster than paid clicks, which means the new inventory is mostly early-stage questions sold through the same auction at the same rising prices.
That changes who carries the risk. In 2016, cheaper clicks made experimentation affordable. In 2026, the advertiser pays full price to find out whether discovery traffic turns into revenue.
The numbers you cannot see yet
Google does publish some figures. AI Mode queries are on average three times longer than traditional searches. More than one in six AI Mode searches in the US are multimodal, and Pichai later said one in five globally use voice, images or documents.
The number that matters most is missing. Alphabet does not break out paid clicks or cost per click for AI Mode. The only performance claim is that click-through rates are "already similar" between sponsored and organic product listings inside AI Mode, based on two weeks of Google's own data.
Two weeks, measured by the seller, comparing ads to organic results rather than to classic Shopping ads. That is a hypothesis, not a benchmark. Your own accounts have to confirm it.
What I would change in the account
The first change is structural. Separate upper-funnel spend from conversion spend so they stop sharing one ROAS target. If AI Mode inventory and broad discovery queries sit in the same campaign as branded and high-intent terms, the cheap conversions subsidise the expensive exploration and nobody can tell what works. I made a version of this argument in your ROAS is lying to you now.
The second change is in how you judge discovery. Upper-funnel queries rarely convert in the same session. Judge them on what they create later: branded search lift, returning visitors, assisted conversions across a longer window. Agree on that window with finance before the spend starts, not after the results disappoint.
The third change is a holdout. Run a geographic or time-based test where one segment gets AI Mode and broad discovery exposure and another does not. If revenue does not move, you have your answer, and it cost you a test budget instead of a full year of spend.
Google is also launching multi-turn ads that can appear in follow-up questions. That means your ad might appear three questions into someone's research, long before they compare prices. Creative written for the last click will look strange there. It needs to answer a question, not shout an offer.
The burden of proof moved
None of this means AI Mode is bad inventory. Google may be right that these queries are incremental and will monetize over time. Longer, more specific questions carry real intent signals, and the advertisers who learn to read them early will pay less to learn than everyone else.
But the commercial structure is clear. Google gets paid at auction prices from day one. The advertiser carries the uncertainty about whether early-stage attention converts.
For brands that need a team to rebuild accounts around this split, it is exactly the kind of work difrnt runs for mid-market and enterprise clients. Whoever runs it, the principle is the same.
The queries moved up the funnel. Make sure your measurement moved with them.