News and Publisher Visibility in Microsoft Copilot

Microsoft Copilot surfaces news the way Bing sees it: articles crawled by Bingbot, ranked by Bing News, and summarized into answers with numbered citations back to the publisher. Uniquely among AI assistants, Copilot gives publishers a documented, granular opt-out — Bing's September 2023 NOCACHE and NOARCHIVE controls — which turns the usual all-or-nothing AI dilemma into an actual policy decision. Visibility strategy starts with making that decision deliberately, then optimizing the pipeline you have chosen to stay in.

The publisher control matrix

These are meta robots tags, applied per page — which means the decision can be editorial, not sitewide. Some publishers keep commodity coverage open for citation reach while tagging premium investigations NOCACHE, preserving the link economy on their most expensive journalism.

Freshness: the Bing News pipeline

Copilot's news answers reward whoever Bing crawled most recently with the most extractable account. Register with Bing's PubHub to establish news-source status, submit news sitemaps, and adopt IndexNow so updates reach Bing within minutes — for developing stories, the outlet whose third update is indexed first often becomes the cited source for the whole cycle. Update articles in place with honest timestamps; Bing's freshness systems reward genuine revision and are unimpressed by republishing tricks.

NewsArticle schema and the paywall declaration

Mark every story with NewsArticle JSON-LD: headline, datePublished, dateModified, author with a linkable profile, and publisher. For paywalled content, declare it — isAccessibleForFree: false with a hasPart block identifying the gated section — while deciding what Bingbot may read. Full-text access for the crawler makes your reporting quotable in answers; truncated access limits you to headline competition. Neither choice is wrong, but incoherent combinations (blocking the crawler while expecting citations) are.

Writing for the citation, not just the click

Copilot compresses articles into two-sentence summaries, so the attribution goes to whichever source stated the key fact most extractably. Front-load the lede with the who-what-when and the specific number; attribute exclusives explicitly ("documents obtained by...") because models propagate attribution language into their answers. Structured sidebars — timelines, "what we know" blocks, FAQ sections on running stories — carry citation value for weeks after the original cycle, functioning as the durable asset under the perishable story.

Defending the referral economy

Measure three things monthly: your citation share on a panel of category news prompts, attribution position within answers, and referral sessions from bing.com and copilot.microsoft.com. When citation share drops, the cited-source lists reveal who took it and with what content. Menra's citation tracking automates the panel and maps which URLs carry your visibility; pair it with the measurement fundamentals in tracking AI mentions. Publishers who instrument this now are the ones negotiating from data — with platforms, and with their own paywall strategy — instead of anecdote.