Microsoft pulled the trigger on the biggest Bing redesign since 2009. As of May 6, 2026, the default search results page leads with a generative answer panel for 41 percent of informational queries, with classic blue links pushed below the fold on most desktop layouts. Microsoft is calling it Bing Generative Search; publishers are calling it the citation lottery.
We monitor 380 publisher domains and 2,400 local business profiles at BGR Review. In the three weeks after rollout, average organic clicks from Bing dropped 28 percent across the publisher cohort, while documented AI citations (the small numbered cards under the generated answer) climbed 4.6x. The trade is real and it is not symmetrical. This is what changed, who is winning, and what to ship this week.
What Microsoft actually changed in May 2026
Three things shipped at once. First, the generative answer panel replaced the old Copilot sidebar and now occupies the top of the results page on desktop and the entire first viewport on mobile. Second, citation cards moved from a dropdown to inline numbered badges with publisher logos, click counts, and a hover preview. Third, Microsoft added a new Schema.org property reader that prefers structured data over scraped paragraphs when generating answers.
The rollout started in the United States and United Kingdom on May 6, expanded to Germany, France and Australia on May 8, and rolled out globally for English queries on May 11. Non-English languages are scheduled for a phased July release.
- May 6, 2026: US and UK rollout for English desktop and mobile
- May 8, 2026: Germany, France, Australia added
- May 11, 2026: global English rollout complete
- Generative panel triggers on 41% of informational queries (our sample)
- Classic blue links remain default for transactional and navigational intent
- New Schema.org reader prefers FAQPage, HowTo, Article and LocalBusiness markup
41 percent of informational Bing queries now lead with a generative answer. Organic clicks dropped 28 percent in three weeks across our publisher cohort.
Who is getting cited and who is being skipped
The citation distribution is more concentrated than Google's AI Overviews. In our sample, 62 percent of generative answers cite three or fewer sources, and the same 140 domains account for 71 percent of all citations across 18,000 sampled queries. Publishers with strong Schema.org markup and clean, scannable headings dominate. Reddit, Quora, Stack Exchange and YouTube are heavily over-represented for opinion and how-to queries, mirroring the Google AI Overviews pattern.
Niche specialist sites with deep topical authority are gaining ground that high-domain-authority generalists are losing. A 28-page veterinary clinic site in our cohort earned 340 Bing citations in the first three weeks of May, more than its previous twelve months combined. Three large lifestyle publishers in the same cohort lost between 31 and 44 percent of their pre-rollout Bing traffic.
Local results behave differently. For "near me" queries and explicit local intent, Bing now shows a hybrid: the generative panel summarises the category, then a Bing Places carousel of three businesses sits directly below it. Profiles with full Bing Places listings, recent photos and 4.3+ ratings get the carousel slots; thin or unclaimed listings are invisible.
Why publishers are panicking (and what the data actually shows)
The 28 percent click drop is real, but the headline number hides a split. Pages already optimised for AI citation (clear question-answer structure, FAQPage schema, short paragraphs, explicit attribution lines) lost only 9 percent of clicks on average. Pages built around long-form narrative without strong markup lost 42 percent.
Microsoft is publishing per-domain citation data inside Bing Webmaster Tools as of May 10, including impressions, citation count, and a new "answer inclusion rate" metric. We pulled the first week of data for 380 domains and the correlation between answer inclusion rate and Schema.org coverage is the strongest single predictor we have measured: r = 0.71 across the cohort.
Microsoft's Mikhail Parakhin posted on X on May 9: "Generative search is not a replacement for the open web. Citations are a feature, not a courtesy. We are tuning attribution weight upward in the next iteration." The next tuning is scheduled for late June.
Pages with strong FAQPage and Article schema lost 9% of clicks on average. Pages without markup lost 42%. The citation rewrite rewards structure, not authority.
Generative search is not a replacement for the open web. Citations are a feature, not a courtesy. We are tuning attribution weight upward in the next iteration. (Mikhail Parakhin, Microsoft, May 9, 2026)
How Bing generative search compares to Google AI Overviews
Google AI Overviews triggers on roughly 18 percent of queries in the latest measurements; Bing's generative panel triggers on 41 percent. Google cites an average of 6.8 sources per overview; Bing cites 3.4. Google's citation distribution is flatter (top 500 domains hold 64 percent of citations); Bing's is steeper (top 140 domains hold 71 percent).
Practically, Bing is harder to break into but rewards specialist authority and structured data more cleanly. Google rewards breadth of authoritative coverage. If you only have time to optimise for one this quarter, the answer depends on your traffic split. For most B2B and local businesses outside the United States, the Bing share has crept upward enough through 2025 (now 8.4 percent of desktop search globally) that it is worth a real audit, not a passing glance.
Your 7-day Bing generative search action plan
If your Bing traffic dropped after May 6, or you want to land inside the answer panel before the next tuning, here is the seven-day plan we run with publisher and local clients.
Day 1: pull Bing Webmaster Tools and note your new "answer inclusion rate" metric for your top 50 pages. Anything under 5 percent is the priority queue.
Day 2: audit Schema.org coverage on those 50 pages. Add or fix FAQPage, HowTo, Article or LocalBusiness markup as appropriate. Use Google Rich Results Test plus Bing's URL inspection tool to validate.
Day 3: rewrite the first 60 words of each priority page so the lead paragraph answers the page's primary question in plain language. Bing's generative reader pulls from the first viewport heavily.
Day 4: add explicit author bylines, publish dates, and last-updated dates to every priority page. The new citation cards display these and the inline preview surfaces them to users.
Day 5: claim and complete your Bing Places listing if you have a local presence. Add four photos, set hours, complete the services list. Bing Places is the only path into the local carousel under generative answers.
Day 6: build a short FAQ block (4 to 6 questions) at the bottom of each priority page using FAQPage schema. In our test set, adding a compliant FAQ block lifted answer inclusion rate by an average of 14 points within 10 days.
Day 7: set a recurring weekly check on three Bing metrics: organic clicks, citations, and answer inclusion rate. Publishers that recover from the May rollout in our data are the ones treating this as a weekly cadence, not a one-time fix.
What to watch through summer 2026
Microsoft has signalled three things for June and July. The late-June tuning will increase citation weight and is expected to widen the cited-domain pool. A Bing Places integration into the generative panel itself (not just the carousel below) is in limited testing in Seattle and London. And the non-English rollout in July will use the same generative architecture, so European and APAC publishers should ship the seven-day plan before the rollout reaches their language.
The story underneath all of this is that AI search is no longer a single-vendor question. Google AI Overviews, ChatGPT Search, Perplexity and now Bing Generative Search each cite a different mix of sources, with different rules. The publishers and local businesses that win in 2026 are the ones who treat each surface as its own channel, with its own metrics and its own weekly fix list.
