Skip to content

NEWS ITEM

News

Microsoft makes Web IQ broadly available, adds Scopes to include or exclude sources

Microsoft says Web IQ is now open to all Azure customers at lower prices, with Scopes to include or exclude sources. Its speed and product-match figures are self-reported; no prices are listed.

Key takeaways

  • Microsoft said on Nov. 17, 2026, that Web IQ, its search and grounding service for AI agents, is now available to all Azure customers with new, lower pricing. Neither announcement lists the new prices [1][2].
  • A new feature called Scopes lets Web IQ customers include or exclude sources by domain, path prefix or exact URL. Each scope can hold up to 10,000 included and 10,000 excluded selectors, and exclusions win when rules conflict [2].
  • Microsoft says Web IQ now powers the web search tool in the Foundry Responses API, replacing Grounding with Bing. It also says Bingbot honors robots.txt, and that content from sites that disallow Bingbot is not in the index Web IQ draws on [1][2].

Direct answer: On Nov. 17, 2026, Microsoft made Web IQ, its search service for AI agents, available to all Azure customers at lower list prices. It added Auto and Auto Advanced retrieval modes, Weather and Places data, health-aware ranking and Scopes, which lets agent builders include or exclude sources, plus a Commerce private preview [1][2]. Its speed, quality and product-match claims are Microsoft’s own, and the posts cite no independent test [1][2].

What happened: In a Bing Blog post signed by Jordi Ribas, President, Search & AI, Microsoft said Web IQ “is available to all Azure customers with new, lower pricing” [1]. A companion Microsoft Tech Community post for developers, published the same day, describes what shipped, how each piece works and the limits developers should know [2]. Microsoft launched Web IQ on June 2, 2026, describing it as “a search engine for AI systems” built on the Bing global index [3]. This article describes the announcements as of Nov. 19, 2026.

What did Microsoft ship?

According to the developer post, Auto is now “the new default way to call Web IQ.” One request fans out across Web, News, Images, Videos, Local and Weather and “returns only cited passages” [2]. Auto Advanced runs additional retrieval passes: it checks whether the evidence gathered is sufficient, rewrites the query to fill gaps and retrieves again before returning cited evidence [2]. The developer post lists Auto, Auto Advanced, Weather, Places and health-aware ranking as generally available. It lists Commerce as a private preview, and Scopes as generally available on Web and News but in preview on Auto [2].

Microsoft says Places returns structured information on more than 207 million businesses, landmarks, attractions and service-area businesses across more than 200 markets [2]. Commerce draws on “more than 40 billion offers” and includes a Commerce Search API and a Commerce Offers API, both in private preview [1][2]. The Bing post says that, in a test of 500 representative product-category queries, “Microsoft’s solution returned 57% more matched products than the next closest competitor”; its footnote describes this as a Microsoft internal assessment and does not name the competitor [1]. For health questions, Microsoft says Web IQ applies health-aware ranking using source-quality signals informed by principles published by the National Academy of Medicine. It also says the service is not intended for individualized medical advice [1][2].

The Bing post’s subheading reads “Powering frontier AI labs, trusted by enterprises,” but neither announcement names an AI lab that uses Web IQ [1][2]. Customers quoted in the Bing post are Genspark, Publicis Groupe and National Grid. Genspark’s co-founder is quoted as saying Web IQ cut its voice assistant’s response time “from 6 seconds to 0.6 seconds” [1]. Microsoft also says Web IQ offers “industry-leading p95 latency” and has open-sourced a FreshSearch freshness benchmark that readers can replicate [1]. These are Microsoft’s and its customer’s own claims.

What does Scopes change for which sites agents use?

Scopes lets a Web IQ customer create a named list of included and excluded sources and pass it with each request, so the restriction applies at retrieval rather than after results come back [2]. A selector can be a domain (which also matches its subdomains), a path prefix or an exact https URL. Wildcards, ports and query strings are not supported, and changes “typically propagate in under 30 minutes, with no SLA” [2]. Microsoft gives examples such as grounding only on approved sources in a regulated workflow, applying a tenant-wide block list, or searching only a company’s own sites [2]. The developer post also describes Premium Content, which lets customers add Microsoft’s catalogue of premium publishers or their own licensed data to the same ranked response [2].

What does Microsoft say about publishers?

The Bing post describes Bingbot as Microsoft’s documented crawler “supporting Bing Search, Copilot, and Web IQ” [1]. It lists three publisher mechanisms. Robots.txt governs crawling, and “Bingbot does not crawl sites that publishers disallow, so that content is not in the index Web IQ draws on.” Page-level AI grounding preferences announced on Oct. 30 cover grounding. NoIndex keeps pages out of the search index [1]. The developer post says Web IQ now powers the web search tool in the Foundry Responses API, replacing Grounding with Bing, with broader integration across Foundry experiences rolling out in 2027 [2]. AVJ covered Microsoft’s Oct. 30 announcement of that grounding preference.

Why this matters for AI visibility

Our view: When a scope is applied, a business building an agent on Web IQ can restrict which sites that agent retrieves from. Microsoft also lets customers fall back to an unscoped search if a scope cannot be applied [2]. That adds a customer-defined layer on top of the Bing index and publisher controls. The posts do not say whether site owners can see when a customer has excluded them. Microsoft’s posts give no usage figures for Scopes or for how often it excludes sources, and beyond the customers quoted they give no breakdown of who uses Web IQ. Treat the performance and product-match figures as Microsoft’s claims until independent tests exist.

What we don’t know yet

  • The new prices. Both posts say pricing is lower but list no figures [1][2].
  • Which frontier AI labs use Web IQ. The Bing post’s subheading refers to “frontier AI labs” but names none; the customers it quotes are Genspark, Publicis Groupe and National Grid [1].
  • How much of Web IQ traffic uses Scopes, and whether site owners can learn when a customer has excluded them. The posts do not say [1][2].
  • How Web IQ applies the Oct. 30 page-level grounding preferences in practice, beyond Microsoft’s statement that publishers can set them [1][2].
  • Independent results for Microsoft’s latency, freshness and product-match claims. Microsoft has published code for its FreshSearch benchmark, but the posts cite no outside test [1].

What to do now

  1. Check that robots.txt and any bot rules let Bingbot reach the pages you want available to Web IQ. Microsoft says Bingbot does not crawl sites that disallow it, so that content is not in the index Web IQ draws on [1].
  2. Review the page-level AI grounding preferences Microsoft announced on Oct. 30, and decide which pages, if any, should opt out of grounding [1]. Our decision framework for AI crawlers and AI search controls sets out the trade-offs.
  3. If you build agents on Microsoft’s Foundry Responses API, note that its web search tool now runs on Web IQ instead of Grounding with Bing, and retest your grounding results [2].
  4. Do not cite Microsoft’s 57% product-match figure or latency claims as independent evidence. The product-match figure comes from a Microsoft internal assessment that does not name the competitor [1].

Disclosure: AI Visibility Journal uses Microsoft’s Copilot API as an ordinary user. It has received no promotions, credits or other incentives from Microsoft, and has no sponsorship or special relationship with Microsoft.

Update log: Nov. 19, 2026 — First published.

Sources

[1] Ribas, J. (2026, Nov. 17). “Web IQ is now broadly available: Build better agents that keep up with the world.” Bing Blog. Accessed Nov. 19, 2026. https://blogs.bing.com/webiq/2026/11/web-iq-is-now-broadly-available-build-better-agents-that-keep-up-with-the-world/

[2] Microsoft Tech Community, Azure Dev Community Blog. (2026, Nov. 17). “What’s new with Web IQ: Ignite 2026 edition.” Accessed Nov. 19, 2026. https://techcommunity.microsoft.com/blog/azuredevcommunityblog/whats-new-with-web-iq-ignite-2026-edition/4574183

[3] Microsoft Bing. (2026, June 2). “Announcing Microsoft Web IQ.” Bing Blog. Accessed Nov. 19, 2026. https://blogs.bing.com/webiq/2026/6/announcing-microsoft-web-iq/

How this was made

Prepared with AI assistance from the Microsoft announcements cited above, then edited by Nhi Phan. Performance and product-match figures are Microsoft’s own and are attributed as such. Any change will be recorded in the update log and in our Corrections Policy and Log.

More from AVJ

Get the next one in your inbox.

The Visibility Brief: weekly, evidence-led, free. Every Friday.