Microsoft Advertising this week confirmed that Predictive Matching, a Search campaign feature that has existed alongside traditional keyword targeting, is being retired and folded into AI Max, the company's broader artificial intelligence layer for search advertising.

The change appeared as a notice inserted into several Microsoft Advertising help documents, including the page explaining keyword match types. Rather than issuing a standalone press release, Microsoft chose to embed the update directly into existing support content, a method the company has used before for feature transitions that do not warrant a dedicated announcement. The notice reads, in full: "Predictive Matching is moving to Search term matching within AI Max. AI Max builds on existing matching capabilities with enhanced AI-powered intent understanding, improved transparency, and additional optimization features. No action is required at this time. Updates will be shared as they become available," according to Microsoft's help documentation.

No effective date accompanies the notice. Microsoft has not said whether the transition is already underway for some accounts, scheduled for a future date, or being rolled out gradually across regions and account types. The absence of a firm timeline places this announcement in a category increasingly familiar to advertisers who track Microsoft's Search platform: substantive changes disclosed through documentation updates rather than through blog posts, press briefings, or the LinkedIn channel the company's product liaisons typically use.

What Predictive Matching did

To understand what is being retired, it helps to understand what Predictive Matching was designed to do, and why it sat somewhat apart from the rest of Microsoft's keyword system.

Microsoft Advertising's keyword match types - broad match, phrase match, and exact match - determine how closely a search query must resemble an advertiser's chosen keyword before an ad becomes eligible to show. Broad match is the most permissive of the three, matching not only the keyword itself but synonyms, related searches, and queries that share no literal terms with the keyword at all. Phrase match sits in the middle, requiring that a query's meaning align with the keyword or a meaningful part of it. Exact match is the narrowest, requiring the query to carry the same meaning as the keyword, though Microsoft's system does still account for spelling variants, plurals, and closely related phrasing under what it calls close variants.

Predictive Matching operated differently. Rather than adjusting how strictly a query needed to match a keyword, it functioned as a supplementary layer sitting on top of an advertiser's existing keyword list. According to Microsoft's documentation, when enabled, Predictive Matching drew on four inputs - an advertiser's final URL, ad titles, ad text, and existing keywords - to identify new, highly relevant search queries that the keyword list alone might not have captured. The stated purpose was to reduce the manual work of mining search query reports for missed opportunities. Microsoft's own phrasing framed it as a way for advertisers to stop analyzing search term reports to find new keywords, since the system would instead surface relevant queries automatically.

The feature carried two hard constraints that limited its reach from the outset. First, enabling Predictive Matching automatically applied broad match to every keyword in the campaign, and advertisers lost the ability to set match types on individual keywords once the setting was turned on. Second, Predictive Matching only worked alongside conversion-based automated bidding strategies: Maximize Conversions, Target CPA, or Target ROAS. Portfolio bid strategies were explicitly excluded. Microsoft's documentation also noted that access to the feature was uneven across the advertiser base, telling users who lacked it that it was "coming soon," language that remains in the current help pages even as the retirement notice sits beside it.

Enabling the setting required navigating either the campaign creation wizard, where advertisers select a Search campaign type tied to Drive Conversions or Generate Leads goals before reaching an Advanced Campaign Settings screen with a Predictive Matching checkbox, or the settings panel of an existing campaign under the same checkbox. Disabling it followed the identical path in reverse.

Where predictive targeting fits in

Microsoft's documentation draws a firm distinction that is easy to overlook: Predictive Matching applied to Search campaigns, while a separately named feature, predictive targeting, applies to Audience campaigns. The two are not interchangeable, despite the similarity in naming, and Microsoft's help material is explicit that the retirement notice concerns Search campaigns and the matching feature specifically, not the audience-side targeting tool.

Predictive targeting, according to Microsoft, helps advertisers find new audiences they may not have already considered, using a combination of ad content, landing pages, and audience intelligence signals to identify potential customers likely to convert. That feature does not appear to be affected by the notice covering Predictive Matching, and nothing in Microsoft's current documentation suggests the two tools are being merged or discontinued together.

AI Max as the successor architecture

Predictive Matching's disappearance is not being presented as a subtraction from the platform. Microsoft frames it as an absorption: the underlying job the feature performed - surfacing relevant queries beyond an advertiser's literal keyword list - continues, but inside AI Max rather than as a standalone setting.

AI Max is Microsoft's umbrella term for a set of artificial intelligence capabilities layered on top of Search campaigns, and its scope has been expanding steadily over the past several months. According to Microsoft's own characterization of the system, published as part of a broader set of AI advertising tools on June 17, 2026, AI Max in Search campaigns covers three mechanisms: query matching, which determines which search terms an ad becomes eligible for; asset personalization, which determines what the ad shows; and URL routing, which determines where the resulting click lands. Those three mechanisms map closely onto what Predictive Matching used to do on its own - matching, using ad assets as an input, and connecting the result to a landing page - which is consistent with Microsoft's description of AI Max as building on existing matching capabilities rather than replacing them with something unrelated.

That same June 17 announcement disclosed that AI Max was expanding to open pilot over the following month, alongside new controls including brand inclusions and exclusions to shape traffic from branded queries, term exclusions that let text assets omit phrases conflicting with brand guidelines, and search term landing page reporting intended to connect the full path from search query through to landing page outcome in a single report. Whether today's Predictive Matching retirement notice is a direct extension of that June pilot expansion, or a separate step in the same broader consolidation, Microsoft's documentation does not specify.

The language Microsoft uses to describe the change is notably close to language it has used elsewhere. The phrase "enhanced AI-powered intent understanding" echoes descriptions the company has applied to other AI Max components, and "improved transparency" mirrors stated goals behind reporting features Microsoft has introduced across Performance Max and AI Max over recent quarters. That consistency suggests the Predictive Matching move is being positioned internally as part of the same initiative rather than a discrete, unrelated retirement.

What advertisers cannot yet see

Several operational questions remain unanswered in the current documentation. Microsoft's notice specifies that "no action is required at this time," language that typically signals an automatic migration rather than one requiring advertiser configuration, but the company has not confirmed whether campaigns currently using Predictive Matching will be moved automatically into equivalent AI Max settings, whether advertisers will need to opt in separately, or whether there will be a transition period during which both systems operate in parallel.

Bidding strategy compatibility is another open question. Predictive Matching required Maximize Conversions, Target CPA, or Target ROAS, explicitly excluding portfolio bid strategies. Whether the successor functionality inside AI Max carries the same restriction, a different one, or none at all, is not addressed in the current notice.

The uneven rollout that characterized Predictive Matching itself compounds the uncertainty. Because not every advertiser had access to the feature being retired, the population affected by its disappearance is not the full base of Microsoft Advertising Search campaign users, but a subset that had already been granted the setting, layered on top of whatever subset gains access to AI Max's expanding query matching capability as it moves through its own staged rollout.

Industry pattern: match-type erosion across platforms

Microsoft's move sits inside a wider pattern that has been building across search advertising platforms for more than a year, and understanding that pattern gives the announcement more weight than it might carry in isolation.

Google has pursued a parallel trajectory with its own AI Max product, introduced as a distinct match type inside Google Ads. Industry analysis of that rollout found that Google's system prioritizes inferred user intent over literal query text when determining ad eligibility, a design philosophy that produces friction for advertisers accustomed to keyword-level control. Reporting from December 2025 documented advertisers discovering that Google's AI Max was assigning conversion credit to campaigns in ways that made it difficult to separate AI Max-driven traffic from traffic generated by conventional exact and phrase match keywords, since the system's default behavior treated all keywords as broad match regardless of the match type an advertiser had actually specified.

The erosion of exact match's traditional precision has been documented independently of AI Max as well. An analysis published in November 2025 found that an exact match keyword built around the term "hypoallergenic" was triggering ads for search queries such as "allergy sensitive dog food" and "top food for dogs with allergies," queries that excluded the word "hypoallergenic" entirely yet still registered as "Exact match (close variant)" in campaign reporting. That case illustrated a broader shift already underway before either company's AI Max products existed in their current form: search platforms have been moving from text-based keyword matching toward semantic and intent-based systems across all campaign types, not solely within newly branded AI features.

Microsoft's own history with keyword precision includes a countervailing data point. In December 2025, the company clarified that exact match keywords retain absolute priority in its ad auctions, overriding Ad Rank calculations and taking precedence over Performance Max campaigns competing for the same placement, according to Navah Hopkins, Microsoft's Ads Liaison, in a December 18, 2025 social media statement. That auction-level guarantee for exact match keywords is a separate mechanism from Predictive Matching and is not affected by today's notice; the two operate at different layers of the system, one determining which campaign type wins an auction, the other determining which queries a Search campaign's keywords become eligible for in the first place.

Google has also introduced its own version of supplementary keyword discovery through search themes, a feature that lets advertisers supply words and phrases as optimization signals without those signals functioning as literal keywords. Microsoft built a parallel capability inside Performance Max, expanding its search theme limit to 50 per campaign on January 9, 2026, according to Navah Hopkins. The conceptual overlap between search themes and what Predictive Matching used to do - both aim to widen an ad's eligibility beyond a literal keyword list using signals rather than strict text matching - suggests Microsoft has been converging its Performance Max and Search campaign tooling toward a common underlying approach for some time, of which today's notice is one further data point.

Why this matters for advertisers

The retirement of Predictive Matching carries direct operational weight for a specific segment of Microsoft Advertising's user base: Search campaign advertisers who had already enabled the feature and built bidding and budget assumptions around its behavior. Because Predictive Matching forced broad match onto every keyword in an enabled campaign and removed the ability to differentiate match types, advertisers using it had already accepted a tradeoff - broader reach in exchange for reduced control - that mirrors the tradeoff industry analysis has identified in Google's AI Max rollout. Losing the specific mechanism without yet knowing its replacement's constraints leaves that segment of advertisers unable to model forward performance with confidence.

For advertisers who never had access to Predictive Matching, given its uneven rollout, the practical impact of today's notice may be limited to a documentation change rather than an operational one, at least until AI Max's query matching component reaches broader availability. That distinction matters when interpreting how significant this announcement is: it retires a feature that was not universally available in the first place, folding it into a system, AI Max, that is itself still moving through open pilot.

The broader significance lies in what the pattern signals about the direction of automated bidding and keyword targeting generally. Microsoft's notice, combined with parallel developments at Google, indicates that supplementary AI-driven query discovery - tools that sit above or alongside conventional keyword match types - is becoming a permanent layer of search advertising infrastructure rather than an experimental add-on. Advertisers managing Search campaigns across both platforms are increasingly required to understand not just three keyword match types, but a fourth and fifth layer of AI-mediated eligibility logic that determines, in ways not always fully disclosed, which queries an ad can ultimately reach.

Timeline

Summary

Who: Microsoft Advertising, through updated help documentation, notified Search campaign advertisers who had access to the Predictive Matching feature, a subset of the platform's user base given the feature's staged, incomplete rollout.

What: Microsoft is retiring Predictive Matching, a Search campaign feature that supplemented keyword targeting using an advertiser's final URL, ad titles, ad text, and existing keywords to surface additional relevant search queries. The functionality is being folded into Search term matching within AI Max, Microsoft's broader artificial intelligence layer covering query matching, asset personalization, and URL routing for Search campaigns.

When: The notice carries no effective date. Microsoft's documentation states "no action is required at this time" and that updates will be shared as they become available, leaving the timing of the actual transition unconfirmed. The change was first spotted in Microsoft's help pages and reported on July 29, 2026, with the underlying documentation update independently reported on July 31, 2026.

Where: The change applies within Microsoft Advertising's Search campaign infrastructure, distinct from predictive targeting, a similarly named but separate feature that applies to Audience campaigns and is not affected by this notice.

Why: The retirement continues a broader consolidation of Microsoft's AI-driven advertising tools under the AI Max umbrella, a pattern that parallels Google's own AI Max match type and reflects an industry-wide shift from literal keyword-text matching toward semantic and intent-based query eligibility across Search campaigns.