Google Marketing Live 2026: New AI Search Ads Features Explained

Google Marketing Live is the keynote where Google sets the tone for the advertising year, and the 2026 edition had one clear through-line: Gemini is now woven into the fabric of Search ads rather than bolted on the side. The headline announcements all point the same way, from new conversational ad formats to a chatbot that quietly replaces the humble lead form.

As a Google Premier Partner, we sat through the keynote so you do not have to. This is the clear-eyed read rather than the launch-day one: every new Search ads feature, our honest take on it, and exactly how we would put it to work for clients. We have also dropped in Google's own demo clips for the three formats that matter most, so you can see them in motion.

The big picture: Gemini inside the ad, not just behind it

For the last couple of years, AI in Google Ads mostly meant automation under the hood: Smart Bidding, asset generation, Performance Max doing its thing in a black box. GML 2026 moves the AI into the ad unit itself. The creative now adapts to the exact question a person asks, products get explained in real time, and in some cases the ad becomes a back-and-forth conversation rather than a static block of text.

Our take: this is a genuine shift in what an ad is. The unit of work is moving from "write the best headline" to "give Gemini the best raw material and guardrails." That rewards advertisers with strong product data, clear offers and tight feeds, and it punishes thin, neglected accounts. The gap between a well-fed account and a lazy one is about to get a lot wider.

Conversational discovery ads

The first new format tailors creative to the specific question someone is asking. Instead of serving one fixed message, Gemini reads the intent behind the search and surfaces the product features most relevant to that person, inside the search experience.

Our take: this is the format we are most curious to test, and the most demanding to get right. When the ad reshapes itself per query, your job becomes feeding Gemini accurate, well-structured product information and trusting it to pick. That is brilliant when your data is clean and a liability when it is not. For clients: the prep work is feed and content hygiene, not clever ad copy. We would spend the first month making sure every product attribute, benefit and differentiator is captured properly before we let the model improvise on top of it.

Highlighted Answers in AI Mode

Google is also placing high-quality ads as recommendations inside AI Mode's suggestion lists. In the keynote example, a search around learning a language returned a shortlist that included a sponsored app recommendation sitting natively among the AI's own suggestions.

Our take: this is the one to watch for organic-versus-paid tension. As AI Mode becomes the default way a lot of people search, getting recommended inside that answer is the new "ranking first." The line between an AI recommendation and an ad is going to feel blurry to users, which makes relevance and quality signals matter more than raw bids. For clients: we would treat this as a reason to keep SEO and content and paid working off the same brief, because the assets that earn organic trust are the same ones that will qualify a brand for these placements.

AI-powered Shopping ads

Shopping ads can now generate custom product explanations on the fly. Search for something like an espresso machine and Gemini identifies the relevant products, then explains why each one suits what you actually asked for, rather than just showing a price and a thumbnail.

Our take: this is the most obviously commercial of the lot and probably the fastest win for retailers. The explanation is only as good as your product data, so the merchant feed becomes the creative. Rich attributes, real specs and honest descriptions will get rewarded; sparse feeds will get bland, generic blurbs. For clients: this lands neatly on work we already do. We would audit the Merchant Center feed first, fix the gaps, then let the format do the persuading. It is a strong argument for treating your feed as a content asset, not a spreadsheet you set up once and forget.

Business Agent for leads

This is the announcement that made the room sit up. A Gemini-powered chatbot can now sit inside the ad and replace the static lead form. Instead of filling in fields, a prospect asks questions directly, about a university course, a service, a product range, and gets answers in the ad unit before they ever land on your site.

Our take: potentially the highest-impact format here for lead-gen businesses, and the one with the most to get wrong. Replacing a form with a conversation can lift quality and intent, because the people who reach out have already had their questions answered. The risk is the agent answering off-script, quoting the wrong price, or making a promise your sales team cannot keep. For clients: we would not switch this on without a tight knowledge base and a clear set of guardrails behind it. Done properly it could shorten the path from click to qualified enquiry; done lazily it is a brand-safety incident waiting to happen. We would pilot it on one campaign, watch the transcripts closely, and scale only once we trust what it says.

Direct Offers, native checkout and travel

Google is also expanding its Direct Offers pilot. Brands can upload a range of promotions (discounts, giveaways, coupons) and Gemini assembles tailored deal bundles for specific searches. Native checkout, built on the Universal Checkout integration, lets people complete a purchase faster without the usual hop across to a separate basket. And in travel, partners such as Booking and Expedia can surface special offers during AI-assisted trip planning.

Our take: the offers and checkout pieces are quietly significant for ecommerce. Fewer steps between intent and purchase is always good for conversion rate, and letting the model bundle promotions intelligently could squeeze more from the same discount budget. For clients: the watch-out is margin. Automated bundling needs hard rules around minimum margins and stackable offers, or you can discount your way to a great-looking conversion rate and a worse P&L.

The supporting cast: AI Max and Performance Max

Underpinning all of this, Google reiterated its push on AI Max for Search campaigns and AI Max for Shopping campaigns, alongside the now-familiar Performance Max. These are the automation layers that decide where the new formats show and how budget flows between them.

Our take: the more the creative becomes automated, the more the control you have left lives in structure, signals and exclusions. Strong first-party data, clean conversion tracking and sensible campaign structure are what separate an account that uses AI Max well from one that just hands Google the keys. That is unglamorous work, and it is exactly where an experienced team earns its keep.

The stat Google wants you to remember

The number Google leaned on was that 75% of people report making faster, more confident decisions using AI Mode in Search. Take it with the usual pinch of salt that comes with a vendor's own keynote figure, but the direction of travel is real: more searches are being answered inside an AI experience, and ads are following the eyeballs.

What this means for advertisers

Strip away the keynote polish and GML 2026 says three things. First, your data is your creative now: feeds, product attributes and knowledge bases do the persuading, so neglecting them is no longer a minor sin. Second, the ad is becoming an interface, with conversations and tailored answers in place of fixed copy, which raises both the upside and the brand-safety stakes. Third, paid and organic are converging inside AI answers, so the teams cannot keep working from separate briefs.

None of these features are a reason to panic or to rip up what is working. Most are pilots, and the smart move is to test deliberately rather than flip everything on at once. But the accounts that win the next year will be the ones that get their data house in order now, before the formats that depend on it roll out widely.

How we would approach it for clients

Our plan is unglamorous on purpose: audit and fix the Merchant Center feed and product data first, tighten conversion tracking and first-party signals, and build proper knowledge bases before we let any conversational format speak on a brand's behalf. Then we pilot the new formats one at a time, read the transcripts and the numbers, and scale only what earns it. If you want to talk through what Google Marketing Live 2026 means for your account specifically, get in touch and we will give you the honest version.