"AI Max" & The Unified Search Engine: Winning in 2026

On 15 April 2026, Google quietly retired one of the longest-running formats in paid search. Brandon Ervin, Director of Product Management at Google Ads, confirmed that AI Max for Search had moved out of beta and that Dynamic Search Ads, Automatically Created Assets and the campaign-level broad match setting would all auto-upgrade into AI Max from September. Voluntary upgrades begin now. After September, opting out is no longer on the table.

For CMOs and performance leads, this is more than a feature retirement. AI Max stops being a product to evaluate and becomes the default architecture of paid search. By autumn, almost every advertiser will be running it. The 2026 differentiator will be how deliberately each brand governs what runs inside it.

The end of the binary

For most of the last decade, brand and performance lived in different rooms. Brand teams chased reach and equity; performance teams chased conversion. AI Max collapses the wall. With it enabled, a single Search campaign can spend across keyword-anchored intent and AI-generated query expansion, with ads now appearing alongside Google's AI Overviews and AI Mode placements that didn't exist at the start of last year.

Both halves of the architecture only work when each is designed with intent. Manual Search remains the high-intent capture layer; Google's spokesperson confirmed at GA that "keywords remain an essential component of a successful campaign strategy, providing the 'fuel' for our AI." AI Max takes over the catch-all role DSA used to play, surfacing net-new queries the keyword set could never anticipate. 

L'Oréal, an early-access partner, reported twice the conversion rate at a 31% lower cost per conversion after activation, including from queries—like "what is the best cream for facial dark spots?"—that no human-built keyword list would have surfaced.

Data is the new creative director

The variance in real-world results tells a more honest story than the launch decks. Smarter Ecommerce's analysis of more than 250 Search accounts running AI Max, reported by Search Engine Land, found a median revenue uplift of 13%—almost identical to Google's own non-retail benchmark—but a median CPA increase of 16% and a ROAS spread that ran from 42% above baseline to 35% below. "Happiness is expectations minus reality," was Mike Ryan's takeaway from running it. Only around a fifth of campaigns held within their original ROAS target.

‍

The pattern is not random. Account by account, the gap between strong and weak AI Max performance traces back to the quality of the conversion signals the system has to learn from. 

With third-party data continuing to thin, AI Max depends more heavily than its predecessors on first-party inputs: server-side conversion tracking, enhanced conversions, CRM revenue imports and customer-match lists tagged with lifetime-value data. 

Google's own year-end report noted a 14% uplift in conversion signals among advertisers who configured the Google tag gateway—a small change to data plumbing that meaningfully changes what the algorithm can see. The strategist's leverage point has moved out of the campaign manager and into the data infrastructure.

Defending against algorithmic cannibalization

The most expensive failure mode in 2026 is internal. Without explicit brand exclusions and a properly maintained negative-keyword list, AI Max will capture branded queries that the keyword Search campaign would have won at much lower CPC, then bank the conversion. Reported ROAS climbs. Incremental customers do not. The trade press has converged on the same diagnosis: a "Power Pack" or hybrid priority structure that treats AI Max, traditional Search, Demand Gen and Performance Max as distinct functions inside one architecture, rather than competing campaigns inside the same auction.

The meaningful difference between AI Max and Performance Max, and the reason this is recoverable, is that AI Max respects negative keywords at both campaign and ad-group level. Account-level negative lists for branded and navigational queries should be in place before AI Max is switched on, not after. High-margin product lines belong in dedicated tiers with their own ROAS targets, not blended into a single campaign that optimizes toward the average. This is the difference between letting the algorithm grade its own homework and giving it a marking rubric.

From operator to architect

The CMO role inside paid search has changed in degree more than in kind. Where the work used to be configuration—bid adjustments, keyword expansion, match-type management—it is increasingly governance. 

Outside signals reinforce the point:

• AdWeek reported in April that 37% of consumers now begin a search with an AI tool rather than a search engine. 
• BrightEdge data, published by Search Engine Land, showed Google search impressions up 49% in the year following AI Overviews' launch, with click-through rates down close to 30% over the same period. 

‍

Visibility is no longer the same thing as traffic.

The architect's job extends beyond the AI Max settings panel into the wider question of where the brand wants to appear, on whose terms, and against what economic standard. Each automation layer—Google's, Meta's, the major DSPs'—has its own black-box dynamics. The discipline that holds across all of them is the same: define the inputs deliberately, separate the campaign tiers, and measure outcomes against business objectives the platform cannot dictate.

The incremental impact standard

That last point is where the 2026 accountability conversation has settled. Writing for The Drum's year-ahead predictions, Joy Talbot of Analytic Partners argued that success in AI-led search will come from "understanding how AI-led search moments contribute to both brand and performance, not just last clicks." Reported ROAS inside an automated campaign now systematically overstates contribution. Holdout incrementality testing, marketing mix modeling and multi-touch attribution are the rebalancing instruments. Used together, they isolate the conversions the brand would not have earned without the spend, which is the only number that should set next quarter's budget.

‍

{{26-AI-Max-The-Unified-Search-Engine-1="/tables"}}

‍

What this means for advertisers

The work has not got smaller. It has moved. The advertisers who win with AI Max in 2026 are not the ones who automate the most; they are the ones who design the boundaries the automation operates inside, then audit ruthlessly against incremental impact rather than platform-reported credit.

That is the discipline AI Digital exists to operationalize. Through our managed service, we run cross-channel buying with custom KPI optimization across CTV, Search, Display, Social and Native. Through Smart Supply, we curate the inventory layer that AI Max and Performance Max both depend on. Through Elevate, our intelligence platform, we give clients the marketing mix modeling, path-to-conversion analytics and cookieless reach that pull the measurement question back inside the brand's control. If your team is preparing for the September cutover and wants a partner that treats automation as something to be governed rather than deferred to, we should talk.

‍