How Microsoft AI Max for Search Will Reshape Bing and Copilot Ads

Microsoft introduces AI Max for search ads

Microsoft announced AI Max for Search on April 21, 2026, and opened the pilot in May 2026 across Bing and Copilot. [1] The platform is not a single feature but a system built around three connected pillars: expanded query matching for conversational AI searches, structured commerce data through the Universal Commerce Protocol (UCP), and in-platform transactions through Copilot Checkout. For PPC advertisers, the operational question is immediate: do your campaigns, feeds, and creative assets work in an environment where AI agents, not humans, are increasingly making purchasing decisions?

Microsoft frames this around what it calls three eras of the web: the human web (traditional search), the LLM web (AI that synthesizes options), and the agentic web (AI that executes transactions directly). [1] The numbers behind that framing are hard to dismiss. Automated traffic grew 8x faster than human traffic in 2025, AI-driven sessions nearly tripled, and agentic browser traffic is up roughly 8,000% year over year. [1] Whether or not you buy the “three eras” framing, those traffic figures describe a real shift in how queries reach Microsoft’s ad inventory.

Advertisers must opt in to use AI Max. [2] It is not on by default, which gives teams time to evaluate it before committing budget.

How it differs from Google’s AI Max

Google announced its own AI Max for Search campaigns in May 2025 and is now retiring Dynamic Search Ads in favor of it. [11] Both platforms share a name and a general direction: replace rigid keyword matching with AI-driven query interpretation. But the execution differs in ways that matter for campaign strategy.

Google’s AI Max is primarily a campaign-level toggle that expands matching and generates landing page URLs dynamically, building on the existing Search campaign structure that advertisers already know. Microsoft’s version goes further into commerce infrastructure. Copilot Checkout, UCP feed support, and the Offer Highlights ad format are not just matching improvements; they are attempts to make Copilot a transaction layer, not just a discovery layer. [7] Google has not shipped a comparable in-chat checkout product at this scale.

There is also a surface difference. Google’s AI Max operates within Google Search results. Microsoft’s AI Max is designed to work across Bing, Copilot Search, and Copilot Answers, where queries tend to be longer, more conversational, and less amenable to traditional keyword bids. [2] That surface difference is what makes the query matching expansion genuinely necessary rather than optional.

From what I can see in the available data, Microsoft is making a bigger structural bet. Google is improving an existing system. Microsoft is trying to build a new commerce surface from scratch, which carries more risk but also more upside if Copilot adoption continues to grow.

New tools for structured commerce ads

UCP-ready feeds are now supported in Microsoft Merchant Center for businesses in the United States. [12] Microsoft describes UCP as a common language for AI agents to communicate with business systems, and the practical implication is straightforward: if your product data is not structured in a way that AI agents can read and trust, those agents will skip your inventory entirely. [2]

This is where the agentic web argument becomes concrete. Human shoppers browse, compare, and tolerate incomplete product pages. AI agents do not. They evaluate structured data, select based on what is available, and move on if the information is missing or untrustworthy. [1] A product feed that works fine for traditional Shopping ads may fail entirely in an agentic context if it lacks the fields or formatting that UCP requires.

The early results from Shopify merchants suggest the feed quality argument is real. Top Shopify merchants using real-time catalog feeds through the Shopify integration saw nearly 90% growth in impression share within Copilot. [1] That is a large number, and it likely reflects how underserved Copilot inventory was before structured feed support arrived, rather than a steady-state performance lift. Still, it signals that feed quality is already a competitive differentiator on this surface.

Offer Highlights is a related format that lets retailers promote specific offer details, such as free shipping or in-store pickup, directly inside Copilot conversations. [2] Best Buy is among the first brands to activate it, and the format is currently available in English-speaking markets across Copilot, Edge, and Bing. [2] For retailers with strong promotional offers, this format has obvious appeal. For brands without a clear promotional hook, it is less obviously useful.

Enabling purchases directly within Copilot chat

Copilot Checkout is the most commercially aggressive piece of this announcement. It allows users to complete purchases inside Copilot while the merchant stays as the merchant of record, which means the retailer handles fulfillment, returns, and customer data rather than Microsoft. [2] Microsoft is expanding catalog coverage to more than 500,000 merchants and extending Copilot Checkout to its mobile apps. [2]

Keeping merchants as the merchant of record is a deliberate design choice. It avoids the regulatory and liability complexity that comes with Microsoft owning the transaction, and it makes the pitch to retailers much easier: you get a new checkout surface without losing control of your customer relationship. Whether that arrangement holds as Copilot Checkout scales is worth watching, because the data Microsoft collects from in-chat purchase behavior is valuable regardless of who processes the payment.

Microsoft’s Brand Agents, which are AI-powered brand experiences within Copilot, are reporting an average 2x lift in conversions compared to unassisted sessions. [1] That figure comes from Microsoft’s own reporting, so treat it with appropriate skepticism, but the directional signal is consistent with what embedded commerce generally does: fewer steps between decision and purchase means fewer drop-offs. [7]

The bigger question for advertisers is attribution. When a purchase happens inside Copilot, the standard last-click or even data-driven attribution models built around website sessions start to break down. Microsoft has not yet published detailed guidance on how Copilot Checkout conversions flow into campaign reporting, and that gap will create measurement headaches for any team trying to compare ROAS across channels.

Personalizing ads across Bing and Copilot

AI Max for Search uses contextual signals from the conversation itself to match ads to queries, rather than relying on keyword bids or audience lists alone. [5] Creative assets are organized by use case rather than by product, and the system pulls from those asset buckets based on conversation cues. [3] This is a meaningful departure from how most search advertisers currently build campaigns, where the keyword is the organizing principle for both targeting and creative.

Early adopters of AI Max saw a 5% improvement in CTR, and advertisers using Performance Max on the Microsoft platform report an average 8% increase in incremental conversions. [5] Those numbers are modest enough to be credible, and they suggest the matching improvements are real without being transformative on their own.

Copilot queries tend to be longer and more specific than traditional search queries. [2] A user asking Copilot “what is the best cordless vacuum for pet hair under $300 that works on hardwood floors” is giving the system far more signal than someone typing “cordless vacuum” into Bing. AI Max is designed to use that extra signal to serve a more relevant ad, but it also means that ad creative written for short, generic queries may underperform in this context. Advertisers who have invested in detailed, use-case-specific asset libraries will have a structural advantage here.

Microsoft’s Clarity AI Visibility tool is also part of this update, giving advertisers insight into how their brands appear within AI-generated responses. [4] That kind of visibility into AI-surface brand presence is something the industry has been asking for, and it positions Microsoft slightly ahead of where Google currently is on AI response transparency for advertisers.

What this means for PPC workflows

The practical workflow changes from AI Max are concentrated in three areas: feed management, creative structure, and measurement. Feed quality is no longer just a Shopping ads problem. If UCP-ready feeds determine whether your products appear in Copilot’s agentic responses, then feed management becomes a responsibility that touches search campaigns too, not just the e-commerce or merchandising team. [12]

On the creative side, organizing assets by use case rather than by product requires a different briefing process. Most search teams build ad copy around keywords and match types. AI Max asks them to think about the intent and context of a conversation instead, which is closer to how content teams work than how traditional PPC teams work. That is not a criticism of the approach; it is a genuine skills gap that teams will need to close.

Measurement is the hardest part. Copilot Checkout conversions, Brand Agent interactions, and Offer Highlights engagements all represent new touchpoints that do not map cleanly onto existing attribution frameworks. [10] Before scaling budget into AI Max, advertisers should define how they will track and value these interactions, because the default campaign reports are unlikely to capture the full picture from day one.

What to watch in the next six months: how quickly Microsoft expands UCP feed support beyond the US, whether Copilot Checkout attribution gets integrated into standard campaign reporting, and whether the 5% CTR lift from AI Max holds as more advertisers opt in and the system becomes more competitive. The pilot phase always shows better numbers than steady state. The more interesting data will come once AI Max is no longer the exception but the default for Microsoft search campaigns.

Why 77% of Shoppers Use AI for Research But Not for Checkout

How the April 2026 Core Update Shifted 80% of Top Search Results