Microsoft Advertising has begun rolling out Excluded content terms, a control that stops advertisements from serving next to page titles containing words an advertiser has banned. The setting, documented on August 6, 2026, caps each account at 1,000 terms and applies them across campaigns automatically.

The feature sits inside the Content suitability section of the Microsoft Advertising Platform, alongside two older controls. It was spotted by Hana Kobzová at PPC News Feed, who posted about it on X. According to her description, the setting allows advertisers to "prevent ads from displaying alongside specific page titles by adding unwanted terms at either the account or campaign level."

That last detail carries operational weight. According to Microsoft Advertising's help documentation, terms added at campaign level do not stay there: any terms added to a campaign are automatically applied to all campaigns in the account. A control presented as campaign-scoped therefore behaves as an account-scoped one the moment it is used.

Three controls, three different mechanisms

Microsoft now documents three separate content suitability controls, and the documentation is explicit that they operate on different objects. Website control lists prevent advertisements from showing on named domains. Content exclusions prevent them from appearing alongside categories of sensitive subject matter. Excluded content terms block them from showing next to unwanted terms, which the documentation defines as page titles.

The distinction between a domain block, a category block, and a string match on a page title matters for anyone building an exclusion strategy. A domain list catches a publisher wholesale. A category filter catches a topic wherever it appears. A term filter catches one headline on an otherwise acceptable page while leaving the rest of the site eligible.

Naming has not kept pace with the expansion. According to the documentation, content exclusions are called Excluded content themes for some advertisers in the Microsoft Advertising Platform, which places a control named Excluded content themes directly above one named Excluded content terms in the same navigation menu. The two do entirely different things.

Website exclusion lists are applied at the account level in addition to any exclusions set at campaign or ad group level, according to the documentation. Microsoft raised the capacity of those lists to 10,000 websites each, with up to 100 lists per account, an expansion that put centralised placement management into the same Content suitability menu where the new term filter now sits.

What Integral Ad Science blocks before the advertiser does

Underneath the optional controls runs a mandatory layer. According to Microsoft Advertising, the company partners with Integral Ad Science to keep advertisements away from selected content categories, and the partnership blocks by default without any advertiser action. IAS checks websites and Microsoft automatically blocks serving on sites deemed medium and high risk across eight categories: adult, alcohol, gambling, hate speech, illegal downloads, illegal drugs, offensive language, and violence.

That default applies to advertisements serving in audience placements. Microsoft states separately that it completes its own brand safety checks for Microsoft owned and operated sites in addition to the IAS layer, and that it does not serve advertisements with any content it deems unsafe.

The eight-category floor comes with a lineage the industry no longer maintains. According to the documentation, those categories broadly align with the Global Alliance for Responsible Media framework for unsafe content. GARM was dissolved in August 2024 after X filed an antitrust lawsuit alleging coordinated withholding of advertising revenue, and the World Federation of Advertisers settled that suit on July 29, 2026, agreeing that it will not form or restart GARM or a similar initiative. The framework persists inside platform documentation as a technical reference point long after the body that wrote it stopped existing.

Seven optional categories, powered by contextual classification

Advertisers wanting protection beyond the default eight can switch on additional sensitive content categories. According to Microsoft, these are powered by IAS using its Context Control Avoidance solution, the semantic classification suite IAS introduced to read page-level context rather than match keywords.

Seven categories are listed, and the definitions are broader than the labels suggest:

  • Crime covers crime and harmful acts including human rights violations, and the definition explicitly extends to educational, informative, and scientific content on the topic.
  • Infectious diseases and outbreaks covers epidemics and outbreaks, with coronavirus, Zika, and norovirus named as examples.
  • Kids' content covers pages aimed at children and pages aimed at their parents, including parenting forums, travel-with-children content, and children's health.
  • Natural disasters covers negative news or sentiment around hurricanes and tropical storms, plus aftermath such as structural collapses and transit disasters.
  • Politics covers governmental politics.
  • Protests and demonstrations covers protests, demonstrations, brutality, and riots.
  • Sensitive social issues covers immigration, abortion, euthanasia, vaccines, and other inflammatory content.

Three of those definitions absorb material that publishers would classify as ordinary journalism. Educational content about crime, health reporting on outbreaks, and parenting coverage all fall inside categories an advertiser can switch off with a checkbox. The seven-category set therefore functions less as a safety floor than as a news-avoidance dial.

Microsoft attaches a disclaimer to the whole apparatus. According to the documentation, while the controls help advertisers maintain a positive online presence and mitigate risk, the company "cannot guarantee that all content exclusions will be blocked."

Where the feature runs, and where it does not

Content suitability controls are available for Audience, Performance Max, Search, and Shopping campaigns, according to Microsoft. Premium Streaming campaigns are not supported yet. Rollout of the term filter is incomplete: the documentation carries a notice stating that not everyone has the feature yet.

Setup runs through Tools, then Content suitability, then Excluded content terms, then Add, with terms entered one per line before saving. Terms can also be added during Performance Max campaign creation through the targeting settings section. The documentation gives a worked example: an advertiser selling technology hardware who excludes the term faulty tech devices will not have advertisements served next to page titles mentioning faulty tech devices.

The report that surfaced the setting also noted a second Performance Max control already in place: age and gender exclusion at asset group level, with age brackets spanning 18 to 24, 25 to 34, 35 to 49, 50 to 64, 65 and over, and unknown, plus female, male, and unknown for gender.

The overblocking problem the term filter inherits

String matching against page titles is the oldest technique in brand suitability and the one with the best documented failure mode. Keyword blocklists cannot read context, which means a term added to protect a brand can remove inventory the brand had no objection to.

Vodafone narrowed its keyword restrictions on exact match to stop unnecessary overblocking, and the case documented precisely how blunt the mechanism can be: a blocklist entry of die also caught pages containing foodie and diet. That work increased the telecommunications company's news advertising inventory by 10 percent.

Microsoft's implementation raises the same question at account scale. With a 1,000-term ceiling and automatic propagation across every campaign, a single poorly chosen entry travels the whole account without a second confirmation step. The documentation describes that propagation as a time-saving measure.

Why this matters for marketers

The release lands in the middle of a broader repositioning of automated campaign types. Performance Max was sold on the premise that the platform decides placement; the past year has been spent handing advertisers back mechanisms to override those decisions.

On Microsoft's side, placement reports gained conversions, clicks, and spend metrics on May 7, 2026, extending site-level reporting past impressions. The Ad Preview Hub went live inside Performance Max on July 28, 2026, covering MSN, Bing Search, and Outlook. In January 2026, Microsoft made audience ads content targeting generally available with placement and topic options. Each step reduces the amount of the campaign type that operates without advertiser visibility.

There is an auction wrinkle specific to this platform. Microsoft confirmed on December 18, 2025 that Audience advertisements and Performance Max campaigns can compete for the same display placement, with Ad Rank determining which creative serves. An exclusion applied to audience placements therefore has consequences for how a Performance Max campaign competes for the same inventory.

The direction is not confined to Microsoft. Google Ads enabled account-level placement exclusions on January 14, 2026, spanning Performance Max, Demand Gen, YouTube, and Display, replacing per-campaign management. Amazon DSPadded content exclusion categories in December 2025, extending past tier-based filtering into topic-specific blocks. Microsoft had already piloted brand exclusions for Performance Max in October 2024, following brand lists introduced in July 2024.

For publishers, the arithmetic runs the other way. Every term filter switched on at account level removes eligible impressions from pages that match, and page titles are where news sites concentrate their most specific language. A term list built to protect a brand from one category of story will, at 1,000 entries applied automatically across an account, remove a measurable share of news inventory from the auction.

For media buyers, the practical question is auditability. Account-level propagation means the person who adds a term inside one campaign changes delivery everywhere, and the documentation describes no approval layer between entry and application. Agencies running consolidated accounts across multiple clients inherit that behaviour by default.

The underlying limitation has not moved. Placement and term controls operate on named objects, and inventory that declines to name itself accurately stays outside their reach. A term filter reads what a page calls itself. It does not read what the page contains.

Timeline

Summary

Who: Microsoft Advertising released the control. It affects advertisers and agencies running Audience, Performance Max, Search, and Shopping campaigns on the platform, along with publishers whose page titles determine eligibility. Integral Ad Science supplies the classification technology behind the adjacent content exclusion categories. The setting was spotted by Hana Kobzová at PPC News Feed.

What: Excluded content terms is an account-level or campaign-level control that blocks advertisements from serving next to page titles containing specified words, with a maximum of 1,000 terms. Terms added at campaign level are automatically applied to all campaigns in the account. It joins website control lists and content exclusions as the third control in the Content suitability section.

When: The feature was documented in Microsoft Advertising help pages and reported publicly on August 6, 2026. Rollout is incomplete, with the documentation stating that not all advertisers have access yet.

Where: Inside the Microsoft Advertising Platform under Tools, then Content suitability, and during Performance Max campaign creation through targeting settings. Coverage spans Audience, Performance Max, Search, and Shopping campaigns; Premium Streaming campaigns are not supported.

Why: The control adds a page-title layer beneath domain-level and category-level exclusions, giving advertisers a way to block individual pages on otherwise acceptable sites. It arrives as Microsoft, Google, and Amazon each return placement control to advertisers in automated campaign types, and it carries the documented overblocking risk of every string-matching blocklist, amplified by automatic propagation across an entire account.