Picture a shopper in AI Mode. First question: “what’s the best way to get rid of algae in a pool?” Google answers with a few steps. Second question: “is a robot vacuum worth it for a small pool?” Third: “which ones don’t need a separate booster pump?” By the third turn that person is a buyer. They have a pool size, a budget in mind and a short list forming. And as of yesterday, in the U.S., Google can put a paid ad next to that third answer.
Here is the part most accounts will miss. If your Search campaign runs on exact and phrase match, you were probably not eligible for the second or third question. You get the first one, when somebody types something that looks like a product search. The follow-ups, where the decision actually happens, go to whoever let Google’s machines do the matching.
That is the real news inside a short Google post that most people will read as a country list.
What Google switched on
On October 28, Shashi Thakur, who runs Search ads at Google, published two changes. Ads in AI Mode are now scaling in Australia, Canada, India, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, the Philippines and Singapore, with Brazil, Mexico and the U.K. planned for later in Q4. And in the U.S., Google is launching what it calls multiturn ads, “to allow ads in follow-up queries in AI Mode.”
A Google spokesperson put it more plainly to Search Engine Roundtable: “Multiturn ads show up on responses to follow-up user queries as users engage in back and forth conversations with AI Mode.” Search Engine Land confirms the same ten markets and the same U.S.-only follow-up format, and notes there is no launch date for the Q4 countries.
Two numbers came with the post. Google says AI Mode queries are on average three times longer than traditional searches, and that click-through rates are “already similar between sponsored and organic product list items within AI Mode.” The footnote on that second claim points to Google internal U.S. data from August 20 to September 3, 2026. That is two weeks, product listings only, with no actual rates published. Treat it as a sales line.
Why it landed on earnings day
The announcement came the same afternoon as Alphabet’s third-quarter results, and the timing is the motive. Search Engine Journal’s earnings read has Search and other revenue up 12% year over year to $63.35 billion, after 19% in Q1 and 17% in Q2, and essentially flat against last quarter. Chief business officer Philipp Schindler pointed first to currency, the biggest factor, plus retail and media, then said strong AI Mode query growth weighed on revenue because AI Mode monetizes at a lower rate. TechCrunch has the same 12%, 17% and 19% sequence and reports Alphabet shares fell 8% after the results. It also quotes Schindler on the call: “This includes taking our monetization footprint global for the first time.”
Sundar Pichai framed it as patience: “Just like our shift to mobile over a decade ago, we are focused on building the best Search user experience, followed by monetization.” Read that from the buyer’s side. Google has a fast growing surface that earns less than classic Search, the stock dropped on the news, and the fix is more ad slots in more turns in more countries. You are the fix. That does not make the inventory bad. It means Google wants these slots filled more than you need to fill them, and that is a good position to buy from if you go in on purpose.
Who actually gets the follow-up
Google’s help page on ads in AI-powered Search spells out the matching rules, and this is where the decision sits.
Google matches ads in AI Mode two ways. Direct matching is the old way: the query itself is commercial, like “pool vacuum cleaners,” and ads match to it. Contextual matching is new: someone asks a longer, exploratory question, and Google matches ads to “both the user’s query and the AI response.” In other words, Google reads its own answer, sees that it mentioned vacuums and algae kits, and finds you an ad slot.
Then the line that matters. AI Max for Search, broad match, Shopping and Performance Max are eligible for both kinds of matching. Exact and phrase match keywords “aren’t eligible for contextual matching on broader exploratory questions.” Search Engine Journal reads the page the same way.
Follow-up questions in a conversation are almost by definition the exploratory kind. “Which ones don’t need a booster pump?” contains no product name. It only becomes commercial because of everything said above it. So multiturn inventory, in practice, is reserved for the targeting types Google already wants you on. The help page says as much in its best practices list, which tells advertisers to adopt AI Max with search term matching enabled, pair it with Smart Bidding, keep budgets flexible on high-performing campaigns, and switch on text customization and final URL expansion.
Nothing about that is hidden. It is just a different deal than many tightly built accounts think they have. Exact match used to be the way you bought certainty. In a conversation, it buys you the opening line and nothing after it.
What you cannot see
You cannot target AI Mode on its own, and you cannot opt out of it separately from Search. The help page answers both questions with a flat “No.” Existing Search, Shopping and Performance Max campaigns are “automatically eligible.” There is no new campaign, no separate budget, and no extra charge. The controls that carry over are negative keywords, brand exclusions and asset pinning.
Reporting is where it gets thin. Performance in AI Overviews and AI Mode is “included in aggregate” in your standard campaign reports and in the search terms report. There is no AI Mode segment on the page. So a broad match campaign could start picking up third-turn clicks this week and you would see it only as a shift in the blend.
And the search terms themselves are not what the person typed. Google’s ad group prioritization page says that for searches in Lens, AI Mode, AI Overviews and autocomplete, “the search term shown in your reporting represents the best approximation of the user’s intent.” Search Engine Journal flagged that wording in May and pointed out what it breaks: negative keyword mining, compliance review, and every slide that shows executives “what customers are searching for.” The same Google page adds that because these searches are not technically identical to a keyword, your exact match keyword may not get its usual priority. AI picks the ad group by meaning, using all the keywords and landing pages in it.
Put those together for a follow-up turn. You may not know which turn of the conversation your ad showed on. The term you see is Google’s summary of intent, not the shopper’s words. And Google has not said whether a follow-up is matched on that one question or on the whole thread before it. That last one is worth asking your rep in writing, because it decides how much of a person’s earlier conversation your ad is effectively bidding on.
Somebody else writes the line beside your ad
There is one more piece, and it is the one brand and legal teams should read twice. In some AI Mode placements, ads appear with what Google calls AI explainers: “concise, independent text generated by Google’s Gemini models” that introduces a group of sponsored options or highlights context next to an ad. Gemini builds it from the user’s query, the AI response, and “publicly available sources including your website.”
Advertisers “can’t manually write, edit, approve, or opt out of them.” Google says it designs the layouts so people can tell the explainer from your ad copy. Maybe they can. Your customer does not care who wrote the sentence that sits right next to your logo and your price.
So your website just picked up another job. Old comparison pages, a stale returns policy, a spec sheet nobody updated after the last product change: those are now raw material for a sentence about you in a paid placement. Google’s own advice on the help page is to keep website content “clear and up to date” so explainers are accurate. That is the closest thing to an approval step you get.
Canada and the other nine
For readers in the ten new markets, this is less abstract. Your existing campaigns that target those countries became eligible for ads inside AI Mode without anyone touching them. The help page lists English as the only language for ads within AI Overviews and AI Mode, notes Pakistan is text ads only, and lists Indonesia for AI Overviews only.
Here in Canada, that English-only line matters. A Quebec shopper having the same conversation in French is not, by Google’s own page, in this inventory yet. If you run bilingual campaigns, expect the English side to change shape first and do not read a gap between the two as a creative problem.
And the follow-up format is U.S. only for now. Google’s post does not say when, or whether, multiturn ads reach other markets. So a Canadian account gets the first-turn AI Mode slots this fall, while a U.S. account gets the whole conversation. If you run both, do not average them into one “AI Mode” story in your Q4 deck.
If you already read our piece on Google testing AI Mode ads for people who connect Gmail, this is the bigger, quieter version of that move. That was a test on a narrow group. This is the default for every eligible campaign in every market where AI Mode ads run.
What to do this week
Start by writing down your actual position. For each Search campaign that matters, note whether it runs on exact and phrase only, broad match, AI Max, Shopping or Performance Max. That list tells you who in your account is in the conversation and who is only at the front door.
Then make one choice on purpose. If you want to be in the follow-up questions, do not flip your best exact match campaign to broad and hope. Build one fenced campaign with broad match or AI Max on a theme where long research conversations are normal for your category, give it its own budget, set your negatives and brand exclusions before launch, and pin the assets legal cares about. If you decide to stay out, write that down too, so nobody upgrades the account during a Q4 push because a recommendation card said so.
Annotate October 28 in your reporting, and in the Canadian, Australian and other new-market accounts, annotate the day your traffic mix started to move. Because AI Mode is blended into normal reports, the annotation is the only label you will have later.
Read your search terms this week with the right expectation. Longer, question-shaped terms are Google’s approximation, not quotes. Use them for themes and negatives. Stop pasting them into decks as verbatim customer language.
Audit the five pages Gemini is most likely to read about you: your main product or category page, pricing, returns or guarantees, a comparison page if you have one, and your about page. Fix anything stale. In practice that is the only control you have over the explainer.
Send your Google rep three questions in writing. Is a follow-up ad matched on the latest question or on the whole conversation? Can one conversation show and charge for your ad on more than one turn? When will AI Mode appear as its own segment in reporting? Write down whatever comes back, including “not yet.”
And check your budget caps. The help page tells advertisers to keep budgets flexible so the machine can chase both direct and exploratory demand. A cap is how you find out whether the third question is worth paying for before it becomes a standing line in your November plan.
Google needs these slots sold, and it said so on an earnings call. The buyers who do well with that will be the ones who decide which conversation they want to be in, and pay for that one.
