Google Marketing Live 2026: the agentic era begins now

Last week, the latest EMEA edition of Google Marketing Live 2026 was released, and I have the feeling that this edition marks a new curve in the sector. Another one. Every year I say it's the most relevant, but the truth is that this time it seems Google wants to change its board completely. If in 2023 we talked about generative AI as a new layer on existing products; in 2024 and 2025 we saw how it was integrating into Search and Performance Max; by 2026, the discourse is different. Google is going to rebuild its entire advertising stack around Gemini and made it clear from the main stage. Of course, as always, learning will first be in the United States, and we will see how all these new features are incorporated over the months.

"We own the full stack. We don't rent our intelligence."

The central message repeated by Debbie Weinstein (EMEA President) and Steph Davis (VP Customer Solutions EMEA) was directed: the advantage of Gemini is the advantage for your business. The promise is a direct connection from Google's labs (chips, research, models) to customer results. A stack that moves faster than any competitor because it doesn't depend on anyone else. Everyone is losing money in the race, but Google is clear that it will reach the finish line.

For those of us who have been in this since before mobile changed everything (👵🏻), this bet has a very clear reading: they want to turn execution into a commodity. Steph Davis verbalized it bluntly: it's not just about knowing how to set up campaigns, adjust bids, or manage keywords. What differentiates now is the criteria, strategic thinking, the ability to decide what to do and why. That forces the conversation with the client to move. And our internal one too. Does Impact Driven Growth ring a bell? If it doesn't, I recommend you take a look.

Before getting into the nitty-gritty, the figures that support the entire presentation (all official Google):

• 2.5 billion monthly users in AI Overviews
• Queries in AI Mode are 3 times longer than traditional searches
• AI Max for Search comes out of beta with a track record of +27% in conversions
• Google and YouTube are present in 82% of all purchase journeys
• Combining them generates +40% ROAS compared to other media (MMM study with TransUnion)
• 1 billion hours/day of YouTube consumption on televisions
• When a creator talks about your product on YouTube, the viewer is 13 times more likely to search for your brand

The session was structured around four major pillars: new era of Search & e-commerce, performance on YouTube, measurement with ROI verified, and agentic marketing. Let's break it down.

New era of Search & Commerce

The first block was led by Shashi Thakur (VP Search Ads) and Brandon Ervin. The thesis is that the old trade-off between making an intelligent decision and making it quickly has disappeared. Before, making an informed purchase meant opening 15 tabs and praying; today AI Mode returns a personalized recommendation in seconds. And consumer data confirms it: 75% of users feel more confident making decisions with AI Overviews. But the data that really made me raise an eyebrow is another. Once users start using AI Mode, they use ChatGPT less for shopping. It's the first time I've heard Google frame Search as a direct competitor to ChatGPT in purchase intent. For those who doubt the battle being fought, there it is. They have been testing since last year, we'll see how long it takes to become a reality, but AI Max will be the one driving these ads.

AI Max for Search comes out of beta

What was a promise last year is now a live product. AI Max comes out of beta with a +27% in conversions, according to Google, capturing that conversational intent that manual keywords couldn't cover. Long searches like "hotels with views of the Eiffel Tower and breakfast included but cheap" previously found no ads. Now they do.

It's worth reading this +27% with a magnifying glass. On what baseline? In which verticals is it concentrated? And what about the cost per conversion and the quality of the lead that comes in? In accounts where every euro is meticulously controlled, we've seen that the promise of volume sometimes comes with inefficient matchings. The figure is attractive but not inherent. Each business will have to validate this expectation. However, not doing so is not an option, let that be clear.

What's new this year:

• AI Max for Shopping: extends the AI Max framework to standard Shopping campaigns, not just Performance Max. It leverages Merchant Center feeds to match against conversational queries instead of exact product searches.
• AI Max for Travel: specific variant for the travel vertical.

Oliver Kiderle (Product Lead for Search) said it unequivocally during the Q&A: AI Max is going to be the best practice from now on for any advertiser trying to capture all demand in AI-driven search experiences. That means accounts operating solely with restrictive manual keywords are immediate candidates for migration. It's competitive survival.

AI Brief: the control we've been asking for

This is probably the novelty that interests consultancies and agencies the most. One of the historical complaints with PMax and AI Max has been the black box feeling. You launch the budget and hope it works well the first time. AI Brief comes to solve it.

It's a conversational interface within AI Max, powered by Gemini, where you indicate in natural language:

• Messaging guidelines: what you want the ads to say and, most importantly, what you don't want them to say.
• Matching guidelines: what types of searches to prioritize or avoid.
• Audience guidelines: who you want to reach and who you don't.

The big advantage over massive lists of negative keywords or exact matches? With searches becoming conversational and long, the negative model seems to be quickly falling short. AI Brief allows expressing intent in natural language and maintaining control over how the brand is expressed. Brandon Ervin showed the interface live: it even generates previews of how the ads would look before launching them. This is gold for regulated sectors, for clients with very strict brand guidelines, or for contractual conversations where the client needs to see that we maintain control over the message.

Rollout: first in English for AI Max for Search, then Performance Max and AI Max for Shopping.

Direct Offers

Direct Offers enter AI Mode in the United States. They allow offering high-intent users something exclusive at the exact moment of decision. And they're not just classic discounts. They also cover local coupons and personalized offers. The advertiser provides the products and rules, and Gemini handles the individual matching for each shopper. We'll see when it lands in Europe.

AI-Powered Shopping Ads

When a user searches for "a capsule coffee maker that fits on a small countertop and makes iced coffee", AI Mode returns the typical shopping ads but with a considerable upgrade: three curated ads with the product's key features written in natural language, not as a technical sheet. They only activate in PMax and the new AI Max for Shopping campaigns.

Business Agent for Leads

This is the novelty I see the most opportunities for clients with long sales cycles. Within an ad, the user can ask questions and get answers powered by the advertiser's website content. If logged in, their data is pre-filled, generating a much more qualified lead than usual.

Google is testing it with companies in the US, UK, India, and Canada in various verticals. For B2B verticals where the lead arrives cold and cools further as they browse the web, this could be a significant change.

Agentic commerce begins to take shape

The second block was led by Dan Taylor, who heads Google's shopping stack. His opening phrase already sets the tone: "We are entering the era of agentic commerce".

Having an agent shop on behalf of the user requires the entire infrastructure (catalogs, inventories, payments, checkouts) to speak a common language. And here Google has made one of the most strategic announcements of the event.

Universal Commerce Protocol (UCP)

UCP is Google's bet for an industry standard that allows AI agents and merchant systems to communicate with each other without friction. In Europe, there are already strong names operating with it. Zalando and Carrefour are letting their live inventory and loyalty data flow directly into the search result.

Why does it matter for our retail clients? Because if you have a catalog in Merchant Center with clean and updated feeds, you are eligible for all the new layers: Universal Cart, Native Checkout, Direct Offers in AI Mode. If you don't have the feeds right, you're out of the game. This turns feed management quality into a strategic lever, not an operational one. Where before you could survive with a mediocre feed, now you're directly out of the conversation.

Universal Cart

The Universal Cart will arrive in the United States in the coming months, first in Search and the Gemini app this summer, then on YouTube and Gemini. It works as a unified cart between brands and Google surfaces. When you add a product:

• It automatically tracks price changes
• Notifies you when it's back in stock
• Works on Gemini, so it improves as models improve

Native Checkout

Retailers configured with UCP in the United States will be able to complete the purchase within AI Mode with a single click. The user sees the Direct Offer, clicks on Buy now, and the offer is automatically applied at checkout.

Same concept extended to YouTube. Shoppable ads where “Buy now” opens a branded checkout within YouTube with payment via Google Pay. Dan Taylor framed it as reducing the friction between "I want something" and "I bought it". And here is the underlying change: the funnel physically shortens. The video and the conversion are no longer in two separate places, no longer two different metrics, no longer the responsibility of two different teams. This will have organizational implications for any medium-large ecommerce.

Asset Studio rewritten with Gemini Omni, Veo 3.1, and Nano Banana

Google's creative studio is completely redone. Asset Studio now integrates:

• Veo 3.1 for video
• Nano Banana, DeepMind's model for professional photo
• Gemini Omni, the new multimodal model that processes and generates text, image, and video simultaneously and announced in this presentation.
• • Natural language interface: you explain the brief and it generates variations
  • A/B Testing in one click: you choose the most promising assets, click Save and create test, and the system pits them against your current best ads
  • Open API for large advertisers generating thousands of variations

The data justifying the effort is compelling: creativity drives up to 49% of incremental sales. That is, almost half. And the connection with other tools also changes: Asset Studio imports from YouTube Studio, Product Studio, Pomelli, Canva, and Adobe. The question for us is what happens to human creativity in this new scenario? It doesn't disappear, obviously. But the dragon needs to be trained. Just as in 2023 we talked about Midjourney, in 2026 we need to learn to leverage Asset Studio because it's going to be where ads are cooked. At Runroom, we've been doing several tests and its power stands out, especially if what interests you is time to market and not so much having the perfect piece.

Powerful updates are expected for this coming year.

Pomelli for SMEs

Pomelli is the Google Labs tool that makes the most sense for small clients or teams with little creative bandwidth. You paste the website URL and it generates on-brand images and videos instantly, ready to insert into Asset Studio. For clients who don't have their own creative team or budget to produce much, this lowers the barrier significantly.

Performance on YouTube: the platform is no longer just awareness

Third block, led by Melissa Hsieh Nikolic, Director of Product Management for YouTube Ads. The thesis: YouTube is no longer the top layer of the funnel, it's a measurable performance engine that should be integrated alongside Search and PMax.

YouTube Creator Partnerships within Google Ads and DV360

Another important announcement. Google is integrating Creator Partnerships as a unified space within Google Ads and DV360. It allows:

• Search creators by keywords
• See which creators are relevant to your brand
• Contact them and orchestrate sponsored content from the same place
• Measure the partnership metrics

The figure that matters? Working with authentic creator assets generates +20% in conversion lift. YouTube has 3 million creators ready to collaborate with brands.

Boost creator videos directly in the asset picker

Operational detail that will save a lot of manual work. No longer need to first publish the creator's video on a channel and then go find it. From Demand Gen, you can boost the creator's video directly in the asset picker during campaign setup.Cross-campaign reach & frequencyA new beta very interesting for brand advertisers. It allows grouping different campaigns and making them work together towards a unified reach or frequency goal. Without this, historically, massive reach and impact campaigns could overlap for the same impact.YouTube vs Paid socialThe data Melissa put on the table:

YouTube generates 1.3 times higher ROI than paid social

Example: in Germany, 55% of YouTube Shorts viewers are not on TikTok (i.e., different audience, not cannibalization)

Engaged views (more than 5 seconds viewed) versus social scroll: more solid metric to demonstrate impact

Demand Gen as the winning horse

• The data on Demand Gen is the most impressive of the block:
• +30% average growth in conversions in the last 6 months
• Works in any vertical and with any budget

Reaches Google Maps as proprietary inventory, which opens a very powerful new channel for retail with physical stores

New vertical feeds (automotive, for example) and AI-assisted campaign setup

• Adding it to your PMax and Search campaigns yields +10% in ROAS and +12% in sales effectiveness
• Display will be swallowed by this consideration machine starting in 2027.
• Measurement and verified ROI: finally metrics that speak to the CFO
• Fourth block, led by
• Harikesh Nair
• and, in the subsequent panel,

Suzanne Claassen

with Lisa Jakobovits, Jessica Nussbaum and Anjali Raghaavan. The argument is compelling: in an AI era, no innovation matters if your measurement base doesn't capture it. And here Google announced what is probably the most structural change of all GML2026.Two new metrics: ABS and QFCAttributed Branded Searches (ABS). Links the fact that a user sees an ad on YouTube or Demand Gen with AON campaigns, with that same user later searching for your brand on Google. It's the cleanest proof to date that your advertising is generating brand intent, especially useful when the final conversion falls outside the short attribution window.Qualified Future Conversions (QFC)

. Analyzes the key actions the user takes right after seeing your ad and uses AI to causally connect them with future sales up to 6 months later.

• For verticals with long consideration cycles (real estate, financial, B2B, education) this is gold. Finally, we can justify top-of-funnel investment with metrics defensible to the CFO.Together, ABS and QFC offer what Harikesh called speed and truth. Short-term signal and long-term signal in the same dashboard.
• There is still little information, and it seems there will be an associated cost for it to be statistically viable to measure, but it's very good news.Meridian within Google Analytics 360

Meridian, Google's open-source MMM (Marketing Mix Model), enters Google Analytics 360. This is what breaks the silo. For the first time, you can analyze performance on Google, Pinterest, TikTok, Snapchat, Meta, and others in one place, with cross-scenario planning capability. And there's more:Key MMM solutions coming soonA new Incremental Attribution Model in limited alpha this year, which will show incremental results continuously within the GA property

Conversational agentic capabilities so you can literally "talk" to your data in GA

They also announced a new open-source GeoX package as part of Meridian, which can be linked to the MMM for a unified view

• Data Manager and Google Tag Gateway
• The foundation of everything. Data Manager unifies the connection of online, offline, and CRM data. The data:
• Advertisers enriching with first-party data via Data Manager achieve an average +11% incremental ROAS

Google Tag Gateway, Google's first-party tagging solution, delivers an average +14% in conversions

It's worth noting one thing: when Google says data strength, it's not marketing. It's a prerequisite. Without clean and well-connected data, neither generative AI, predictive models, nor Ask Advisor (next block) will perform. If in 2023 we said you had to invest in your own data, in 2026, without it, you simply don't play.

Critical operational detail that cannot go unnoticed. As of June 15, 2026, Google Signals and Ads Consent Mode configurations will be destination-specific. That is:

• Google Signals will control the data you use in Google Analytics
• Ads Consent Mode will control the data you use in Google Ads

The cookies and IDs you collect from the tag or SDK will be separated according to their use. If you decide to set ‘ad_storage’ to ‘denied’ by default in Ads, audiences, bids, and conversions will be affected. For example, remarketing lists may decrease in size.

Before June 15, it's time to audit configurations for all clients

• .
• Agentic marketing: Ask Advisor consolidates everything

The event's closing was led by Steph Davis with one of the most powerful announcements. The consolidation of the entire family of Google agents, which had been in scattered production across different products, into a single advisor: Ask Advisor.What is Ask Advisor

Ask Advisor unifies Google Ads, Google Analytics, Google Marketing Platform, and Merchant Center into a single conversational interface, built on Gemini and with shared memory. The agent remembers the user's goals as they move between tasks.

Examples of questions it would answer:

How do I optimize a campaign for new customer acquisition?Show me abandoned carts of new vs. existing customers, by region

How should I budget for the next quarter? (and does scenario planning)

Status: beta for English accounts, with a broad rollout during the year.

• And that's all for this year. Now we just have to wait to see when all these new features are implemented because some from last year we are still waiting for. We know we will be hearing about these topics for at least the next 6 months, if they don't suffer delays. It seems we will have to start reserving budget to cover all the needs that will arise, whether at a technical, creative, or new functionality level.
• Muéstrame los carritos abandonados de clientes nuevos vs. existentes, por región
• ¿Cómo debería presupuestar para el próximo trimestre? (y hace planificación de escenarios)

Estado: beta para cuentas en inglés, con rollout amplio durante el año.

Y esto ha sido todo por este año. Ahora sólo falta esperar a ver cuándo quedan todas estas novedades implementadas, porque algunas del año pasado todavía seguimos esperando. Sabemos que iremos oyendo hablar de estos temas en los próximos 6 meses como mínimo, si es que no sufren retrasos. Parece que vamos a tener que ir reservando presupuesto para cubrir todas las necesidades que van a ir apareciendo ya sea a nivel técnico, creativo o de nuevas funcionalidades.