Google Marketing Live 2026: What AI Max, Ask Advisor, and the DSA Deadline Mean for Ad Operators

Google Marketing Live 2026 landed on May 20 and the headline is clear: Gemini is now the operating system of Google Ads. Not a feature inside the product — the layer everything else runs on. For operators who manage budgets across Google, Meta, and TikTok, the practical question is what actually changed and what you need to do before September.

AI Max exits beta — and DSA dies in September

AI Max for Search Campaigns is now generally available. The pitch: turn on search term matching, text customization, and final URL expansion together and get 7% more conversions or conversion value at similar CPA/ROAS versus search term matching alone. Hundreds of thousands of advertisers are already on it. Google is citing Lufthansa at +24% ROAS and IKEA at +65% non-branded clicks with +28% incremental ROAS.

The real news is the deadline: starting September 2026, Dynamic Search Ads, automatically created assets (ACA), and campaign-level broad match are force-migrated to AI Max. New DSA campaign creation gets disabled in Google Ads, Editor, and the API from that date. If you are running DSA campaigns today, you do not get to opt out — you get to prepare or get surprised.

New controls that come with AI Max: brand controls (restrict which brand terms can trigger), location controls, text guidelines (constrain how Google rewrites your ad copy), and text disclaimers — a feature aimed at regulated industries that guarantees legally required text appears even when final URL expansion is rewriting landing page targets. That last one matters if you run finance, healthcare, or legal verticals where a misrouted landing page is a compliance event, not just a missed click.

The practitioner concern is real: AI Max has been observed routing granular product queries to generic landing pages instead of matching product pages. That is a regression from what DSA did well. If you are in e-commerce with thousands of SKUs, audit your search terms report weekly after the migration to catch this pattern early.

Smart Bidding Exploration hits PMax and Shopping

Smart Bidding Exploration was already live for Search, where Google claims it delivers 27% more unique converting users. Now it is expanding: PMax beta is live, Shopping campaigns are "a few weeks" behind.

The mechanism: you set a Target ROAS tolerance — how far below your target ROAS you will accept — and Google explores less obvious queries and user segments beyond your current targeting to find incremental conversions. In practice this means Google will spend your budget on audiences and placements you did not explicitly target, and it will do so within a guardrail you define.

If you run PMax with product feeds, this is on by default once it rolls out to your account. Budget monitoring and negative keyword/placement exclusions become more important, not less. The 27% claim is a best-case aggregate; your marginal CPA on the explored traffic will be higher than your core by definition — that is what "tolerance" means.

Paired with this is Demand-Led Pacing, a new feature for Search and Shopping that auto-adjusts daily spend to match consumer demand fluctuations while staying within your monthly budget. If you are used to manual dayparting or intraday bid adjustments, this is Google telling you that layer is theirs now.

Ask Advisor: Gemini across Ads, Analytics, and Merchant Center

Ask Advisor replaces the separate Ads Advisor, Analytics Advisor, and Merchant Center agents with a single Gemini-powered conversational interface. It has a shared memory layer so it retains your goals as you move between products. Currently in beta for English-language accounts globally.

What it can do today: explain performance changes, suggest optimization actions, pull data across Ads and Analytics in one query, and walk through Merchant Center setup. What it cannot do: execute multi-step campaign restructures, handle cross-platform data (Meta, TikTok), or reason about your unit economics beyond what Google surfaces natively.

For operators who manage Google alongside other channels, this is useful but siloed. Ask Advisor knows everything about your Google account and nothing about your Meta CPA or TikTok creative rotation. The gap between a platform-native agent and a cross-channel operating layer is exactly where tools like AdsAgent's MCP integration sit — bridging Google, Meta, and TikTok in a single conversation with full context.

Asset Studio and creative automation

Asset Studio now integrates Gemini, Veo (video), and Nano Banana models, with Gemini Omni coming summer 2026. The headline feature is AI Brief: describe your brand, audience, and guardrails in natural language, and the system generates ad creative with previews. This replaces the manual asset-by-asset creation workflow.

There is also 1-Click Creative Testing — automatically runs new assets against your current top performers — and API access to pull assets from Adobe and Canva. Google is citing that "creative drives nearly half of all incremental ad sales" as the motivation.

The practical takeaway: if you are still manually uploading responsive search ad variations one by one, you are now competing against accounts where Google generates and tests creative automatically. The floor of creative output per campaign just went up.

Other announcements that matter for operators

• Demand Gen to Google Maps — first time ads from Demand Gen campaigns appear on Maps. If you run local/multi-location campaigns, test this early.
• Creator partnership assets in Demand Gen — adding creator-sourced assets to Demand Gen boosts conversion lift by an average of 20%. Note: "only 40% of Demand Gen conversions occur in the first 30 days" — plan your attribution window accordingly.
• Qualified Future Conversions — Gemini-powered metric predicting conversions up to 6 months out. Crew Clothing's YouTube prospecting looked flat at 30 days but showed 70% uplift in long-term conversions. Doc Martens saw 16% revenue jump after feeding first-party data into PMax.
• Shopping Graph now holds 60+ billion listings.
• Meridian (Google's open-source Marketing Mix Model) integrated into GA360.

What to do before September

This is not a "watch and wait" cycle. The DSA migration has a hard deadline and Smart Bidding Exploration is rolling out whether you opt in or not. Here is the practical checklist:

This week:

• Inventory all DSA campaigns and ACA-enabled campaigns. These are the ones that auto-migrate in September.
• Test AI Max on one non-critical campaign now so you understand the controls (brand, location, text guidelines, disclaimers) before the forced migration.
• Check if Smart Bidding Exploration has appeared in your PMax settings. If yes, set an explicit ROAS tolerance — do not let Google pick the default.

This month:

• Audit your search terms report under AI Max. The known failure mode is generic landing pages for specific product queries. Catch it before September scales it across all your campaigns.
• Evaluate your creative pipeline against Asset Studio's AI Brief capabilities. If you are still doing manual RSA variants, the competitive floor just moved.
• If you run Demand Gen, test creator partnership assets — the 20% conversion lift is worth a two-week test budget.

Before September:

• Migrate DSA campaigns to AI Max manually on your own timeline rather than letting Google auto-migrate them. Manual migration lets you set controls and baseline performance before the switch.
• Document your pre-migration CPA/ROAS baselines per campaign. The force-migration will make before/after comparison impossible if you do not have the data saved externally.
• Review Demand-Led Pacing settings when they appear. If your business has predictable demand patterns (seasonal, weekday-heavy), the auto-pacing may conflict with what you know about your own traffic.

The bigger picture: Google is moving from "tools you configure" to "agents you supervise." Every announcement at GML 2026 pushes control further from the operator and closer to the model. That is fine if your business is simple and Google's model understands your unit economics. For operators running cross-channel budgets where Google is one of three or four platforms — and where the optimization target is blended ROAS, not platform-reported ROAS — the gap between what Google's agent knows and what you need to decide keeps widening. That gap is where the next generation of ad operations tooling lives. (For the Meta-side view of attribution trust, see Meta CAPI and iOS in 2026; for cross-platform fundamentals, see why cross-channel attribution breaks at scale.)