I just got back from Google Marketing Live 2026. Two days of keynotes, product demos, and hallway conversations with other brand marketers, agency partners, and Google reps. As someone who manages paid media budgets for a brand — not an agency writing about it for clients — here’s what stood out, what I’m acting on immediately, and what I think the industry is overhyping.
The headline version: Google is no longer selling tools. It’s selling an operating system for marketing, with Gemini as the engine underneath everything. Every single update at GML 2026 was built around AI — not as a feature added on top, but as the default layer that search, shopping, video, measurement, and creative now run through.
That sounds like a press release, so let me be more specific about what it means in practice.
The Big Picture: What Changed This Year
Previous GMLs introduced AI features that worked in isolation. One tool generated ad copy. Another optimized bidding. A third handled audience expansion. They were useful, but disconnected.
GML 2026 was different. Google connected these pieces. Ask Advisor spans Google Ads, GA4, Merchant Center, and Google Marketing Platform in one unified agent. Universal Cart follows a shopper from Search to YouTube to Gmail. AI Brief feeds brand guidelines into every creative the system generates. Meridian now lives inside GA360, not in a separate Python notebook.
The theme wasn’t “here’s a new AI feature.” It was “here’s how all of our AI features now talk to each other.”
For brands, this changes the game in a specific way: the competitive advantage is shifting from campaign management skills to input quality. The brands that feed Google better data — cleaner product feeds, sharper brand guidelines, richer first-party signals — will get better output from an AI system that’s now making more decisions on their behalf.
That shift has real implications for what your team spends time on, how your agency relationship works, and where your next budget dollar goes. I’ll get to all of that.
First, here’s what was actually announced, sorted by how urgently you need to act.
Act Now: These Need Your Attention This Quarter
DSA Is Dying in September. Migrate to AI Max.
This didn’t get a big keynote moment, but it’s the most operationally urgent item from GML 2026. Google is retiring Dynamic Search Ads in September 2026 and folding that functionality into AI Max for Search campaigns.
If your account still runs DSA campaigns — and many do — you need a migration plan now, not in August. AI Max maintains campaign transparency (you keep control over keywords and negative lists) while adding AI-driven creative rotation and audience expansion. It’s a meaningful upgrade over DSA, but the transition takes time to get right.
What I’m doing: auditing every DSA campaign in my account this week, mapping each one to an AI Max structure, and running them in parallel for at least 6 weeks before cutting over. If you wait until August, you’re migrating under deadline pressure with no performance baseline.
AI Brief: Upload Your Brand Guidelines Now
AI Brief was the single most well-received announcement at GML 2026, and for good reason. It directly addresses the biggest complaint brand marketers have had about Performance Max and AI-generated creative: loss of brand control.
AI Brief lets you give Google’s AI a creative brief in plain language — brand voice, target audiences, guardrails, messaging dos and don’ts. The AI interprets those inputs and generates ad creative within those boundaries, with previews you can review before anything goes live.
This changes the relationship between brand teams and Google’s automation. Instead of approving or rejecting individual ad variations, you’re setting the rules the AI follows. It’s a shift from quality control on outputs to quality control on inputs.
What I’m doing: getting my brand guidelines, tone-of-voice documents, and messaging frameworks into a format that Asset Studio can ingest. A sharply written brief produces dramatically better AI output than a vague one. This is worth a day of your team’s time right now.
Product Feed Quality Is Now a Performance Lever
This was the implicit message running underneath multiple announcements — Universal Commerce Protocol, AI-powered Shopping ads, Universal Cart, AI Max for Shopping — and it’s worth making explicit.
Gemini now writes product descriptions dynamically based on what the user searched. It pulls from your feed attributes to generate copy that explains why your product matches their query. If your feed has generic titles, thin descriptions, and missing attributes, Gemini has nothing good to work with.
I talked to several DTC brands at the event who saw meaningful Shopping performance lifts after enriching their feeds with detailed product attributes, lifestyle-oriented descriptions, and complete specification data. One outdoor apparel brand told me their click-through rate on Shopping improved by 30%+ after rewriting product titles to include material, use case, and fit details rather than just brand name and SKU.
What I’m doing: running a full feed audit. Checking every product title, description, image, and attribute for completeness. If your feed is “good enough” for the old keyword-matching Shopping experience, it’s likely not good enough for AI-powered Shopping where Gemini dynamically generates product narratives.
Start Preparing: These Roll Out in Coming Months
Universal Commerce Protocol and Universal Cart
UCP is Google’s open standard that lets product data, pricing, inventory, and loyalty benefits flow between Google surfaces — Search, Maps, YouTube, Gmail, and the Gemini app. The Universal Cart is the consumer-facing result: a persistent shopping cart that follows users across all of Google, with price drop alerts, back-in-stock notifications, and buy-now-pay-later options through Klarna and Affirm.
For ecommerce brands, this is significant. A shopper can discover your product in a YouTube video, add it to their Universal Cart, get a price drop notification in Gmail, and check out without ever visiting your website.
That last part is the tension. Frictionless checkout lifts conversion rates, but it also means Google sits between you and your customer. You get the sale, but you might lose the data-rich on-site experience that powers retargeting, email capture, and lifetime value modeling.
Google is expanding UCP into hotel bookings and local food delivery, which signals this is a long-term infrastructure play, not a pilot they’ll quietly sunset. Brands in ecommerce, travel, and food service should start evaluating UCP integration now.
What I’m doing: evaluating what UCP integration would look like for our product catalog. It’s not live for everyone yet, but early adopters will have an indexing and visibility advantage as Google rolls this out. I’m not rushing, but I am reading the documentation and talking to our Merchant Center rep.
AI Mode Ad Formats: Conversational Discovery Ads and Highlighted Answers
Google introduced two new ad formats specifically for AI Mode, the conversational search experience that’s replacing traditional SERPs for an increasing share of queries.
Conversational Discovery Ads appear within the AI Mode response itself, written by Gemini to match both the user’s query and the surrounding AI-generated content. Instead of a display ad sitting above the organic results, your ad is part of the conversation. Highlighted Answers place your product directly into recommendation lists that AI Mode generates — when someone asks for “best apps for learning Spanish,” your product can appear as one of the curated suggestions.
Google says 75% of users report making faster, more confident decisions when using AI Mode. Whether that number holds up to scrutiny, the directional signal is clear: more search traffic is flowing through AI Mode, and brands need ad presence there.
The interesting strategic question: these formats work with your existing Performance Max, Search, and Shopping campaigns. You don’t need to build new campaigns to appear in AI Mode. But you do need to think about whether your existing ad assets and product data are strong enough to perform well when Gemini rewrites them for a conversational context.
What I’m doing: reviewing my top-performing campaign assets to see how they’d read in a conversational format. If your ad copy is written in traditional “headline + description” mode, it may need to evolve toward more natural, benefit-focused language that works when Gemini weaves it into an AI Mode response.
Asset Studio Gets Multimodal Capabilities
Asset Studio, Google’s creative workspace inside Google Ads, now integrates Gemini Omni — Google’s new multimodal model that can take text, image, and audio inputs and produce video, image, and copy outputs.
In practice, this means you can upload a marketing brief and brand guidelines, and Asset Studio will generate a storyboard, horizontal and vertical video variations, and voiceovers in a single workflow. It connects to Adobe, Canva, and YouTube Studio so your existing assets live in one library.
The one-click A/B testing feature is worth calling out specifically. You can turn any Performance Max asset edit into a structured experiment that measures exact performance lift. Creative testing has always been the thing that teams know they should do more of but rarely find the time for. If one-click testing actually works as advertised, it removes the biggest friction point.
An honest note: Asset Studio was mentioned at GML 2025 too, and the initial launch didn’t live up to expectations. The 2026 version is genuinely more capable — multimodal generation is a step change from basic image variations — but I’d test the quality of generated assets against your brand standards before scaling.
What I’m doing: running a small-scale test with Asset Studio’s new video generation. Uploading one product marketing brief with clear brand guidelines and seeing what comes out. If the quality is usable, this addresses a real production bottleneck for brands that need video creative for Demand Gen and YouTube but don’t have the budget for agency-produced content at scale.
Business Agent for Leads
For B2B and lead-gen advertisers, this is the update worth watching. Business Agent creates an AI agent within your ad that answers prospect questions using content from your website, then collects lead information through a pre-filled form. By the time the lead reaches your CRM, they’ve already engaged with your content and self-qualified.
Google is testing this in education, automotive, and real estate first. If your business has complex purchase decisions and a consultative sales process, this could meaningfully change how top-of-funnel lead gen works.
The risk is quality. AI-generated responses based on your website content will only be as good as your website content. If your site has outdated information, contradictory messaging, or thin product pages, the Business Agent will confidently deliver that bad information to prospects.
What I’m doing: auditing my website content as if it were the training data for a sales agent — because that’s essentially what it is now. Pages that are “good enough for SEO” may not be good enough for an AI that’s representing your brand in real-time conversations.
Watch and Evaluate: These Matter But Aren’t Urgent
Qualified Future Conversions (QFC)
QFC is Google’s attempt to solve the upper-funnel measurement problem. It uses Gemini to predict future conversions based on current engagement signals — branded searches, engaged site visits, video views — up to six months out. The idea: a YouTube campaign that generates no immediate conversions can now be credited with the conversions Google’s models expect it to produce later.
I understand the problem this solves. Demand Gen and brand campaigns have always struggled to justify budget because the conversion window is too short. QFC extends that window with predictive modeling.
But I have reservations. The metric connects current ad engagement to predicted future purchasing behavior using signals that are themselves generated within Google’s ecosystem. Until advertisers can validate QFC predictions against actual outcomes over multiple cycles, this is a metric to monitor, not a metric to optimize against.
QFC is currently in restricted pilot with broader beta access expected later this year. The three new reporting columns — qualified future conversions, cost per qualified future conversion, and qualified future conversion rate — will eventually integrate into bid optimization. That’s when it gets consequential. For now, it’s worth understanding but not worth restructuring your measurement framework around.
Meridian Inside GA360
Meridian is Google’s open-source marketing mix modeling library. At GML 2026, Google announced it will be integrated directly into GA360, so you can run MMM within the same environment as your campaign data instead of exporting to a separate Python environment.
For brands that already run MMM, this is an operational efficiency gain — faster iteration, easier scenario planning, and less data wrangling. For brands that haven’t run MMM before, this lowers the barrier to entry.
The limitation: Meridian in GA360 only works with the paid Analytics 360 tier. Free GA4 users don’t get access. And the bigger issue is that MMM is only as good as the data you feed it. If your non-Google spend data — Meta, TikTok, LinkedIn, CTV, offline — isn’t flowing into GA4 or a connected BigQuery dataset, the model will over-represent Google channels.
What I’m doing: fixing my non-Google spend pipelines first. Making sure Meta, TikTok, and Microsoft Ads spend lands in a connected dataset on a reliable schedule. That’s the prerequisite work that has to happen before Meridian integration becomes useful.
Ask Advisor
Google is consolidating its various in-platform AI agents into one unified experience called Ask Advisor. It works across Google Ads, Merchant Center, GA4, and Google Marketing Platform, retains context across sessions, and can take actions on your behalf — launching campaigns, generating assets, and flagging optimization opportunities.
The practical promise is real: fewer tabs, less switching between tools, and a persistent assistant that can answer cross-platform questions like “which creative is driving the most revenue from new customers in the Northeast?”
The practical risk is also real: an AI that can take actions on your behalf can take the wrong actions on your behalf. The guardrails and approval workflows around Ask Advisor’s action-taking capabilities aren’t fully clear yet.
I’d let this mature before relying on it for anything beyond data queries and surface-level analysis. Use it for questions. Be cautious about letting it make changes.
Demand Gen Updates: Creator Partnerships and Campaign Type Attribution
Two Demand Gen updates worth noting together.
First, Google Ads will now surface relevant creator content that features your brand — including affiliate partnership videos — directly in Demand Gen campaign setup. You can add creator assets to your campaign without leaving the platform. Google’s data shows creator assets increase conversion lift by an average of 20%.
Second, campaign type attribution now shows the conversions Demand Gen specifically contributed to, separately from other campaign types. Previously, Google de-duplicated conversions across campaigns, which made it hard to isolate Demand Gen’s actual impact. This gives you the clearest apples-to-apples view of Demand Gen performance to date.
These two updates together make Demand Gen significantly more measurable and easier to scale with strong creative. If you’ve been skeptical about Demand Gen ROI, the attribution update gives you the data to test that skepticism.
What This Means for Your Team and Your Agency
Here’s the conversation nobody at GML 2026 had on stage, but everyone was having in the hallways.
If Gemini handles creative generation, bidding optimization, audience targeting, and campaign management, what’s left for humans to do? And how should brand marketing teams restructure around this reality?
My take, after two days of watching these demos and talking to other brand-side marketers:
The execution layer is getting automated. Campaign setup, bid management, audience expansion, creative variation, and basic reporting are increasingly handled by AI. Teams that are still primarily staffing for these tasks are going to find diminishing returns on that headcount.
The input layer is becoming the competitive advantage. What separates a brand that gets great output from Google’s AI and one that gets mediocre output? The quality of what you feed it. Better product data. Sharper brand briefs. Cleaner first-party data. Richer conversion signals. Tighter measurement infrastructure.
Strategic direction and quality control remain firmly human. Knowing which markets to prioritize, which audience segments to pursue, which creative angles to test, and when the AI’s recommendations are wrong — that’s the work that creates differentiated value.
For agencies, the implication is stark. The value of an agency that primarily manages campaigns is eroding. The value of an agency that improves your data quality, measurement infrastructure, creative strategy, and feed optimization is increasing. If your agency conversation is still mostly about bid adjustments and keyword management, it’s time to reassess.
For internal teams, this is an opportunity to shift time from repetitive campaign tasks to the strategic and data-quality work that actually moves the needle. The people on your team who understand your customer deeply, who can write a sharp brief, who know what good creative looks like — they’re becoming more important, not less.
My Priority List for the Next 90 Days
After two days at GML 2026, here’s what I’m actually going to do when I get back to the office, in order:
- Audit and migrate all DSA campaigns to AI Max. September deadline is real.
- Upload brand guidelines into Asset Studio through AI Brief. This takes a day and immediately improves every AI-generated creative.
- Run a full product feed audit. Enrich titles, descriptions, attributes, and images. This is now a direct performance lever.
- Audit website content as AI training data. Business Agent for Leads and AI-powered Shopping both use your site content as source material.
- Fix non-Google spend data pipelines. Get Meta, TikTok, and Microsoft Ads spend flowing into GA4 or BigQuery before trying to use Meridian.
- Test Asset Studio’s new video generation on one product campaign to evaluate quality.
- Review Demand Gen campaigns with the new campaign type attribution to finally see real incremental impact.
- Watch QFC and Ask Advisor from the sidelines. Interesting, but not ready for decision-making yet.
The Uncomfortable Truth About GML 2026
Here’s what I kept thinking during the keynote but nobody said out loud.
Google is building a system where the brands that invest in their data, their content, and their measurement infrastructure within Google’s ecosystem will outperform the brands that don’t. Every announcement — UCP, AI Brief, Meridian in GA360, Ask Advisor — deepens the integration between your business data and Google’s AI.
That’s not inherently bad. Better inputs leading to better outputs is how it should work. But it does mean that the cost of doing Google Ads well is no longer just your media spend. It’s the ongoing investment in feed quality, data infrastructure, creative strategy, and measurement sophistication that turns Google’s AI from a generic tool into an effective one.
The brands that treat these as operational costs rather than optional upgrades will have an edge. The brands that try to get by with the same thin feeds, vague briefs, and disconnected measurement they’ve been running for years will find Google’s AI less and less helpful over time — even as it gets more powerful.
That’s the real takeaway from GML 2026. The AI got dramatically better. The question is whether your inputs will keep up.
