- The key findings at a glance
- What gets cited by AI? The short answer from my data
- Which page types get cited by AI?
- Which topics get cited most?
- Do AI citations follow Google rankings?
- Are AI citations concentrated on a few pages?
- Does page age or length matter for AI citations?
- How fast are Bing AI citations growing?
- How to get cited by AI: what this means for your business
- Methods and limitations
- Frequently asked questions
- Want to know if AI cites your business?
My site earned 708,200 Bing AI citations in the three months to late October 2026. In the 60 most-cited pages, comparison pages (X vs Y and alternatives) were missing completely, even though they make up 6.9% of my site. Verb-led how-to pages were cited at 8.8 times their share of the site. And the AI's favourite pages are not Google's: the top 60 Bing-cited pages earned just 0.40% of my Google clicks.
I've been tracking how often Microsoft's AI answers (Bing, Copilot and their partners) cite my website as a source. Bing Webmaster Tools reports this as AI citations, and the numbers have grown fast enough that I wanted to know what sits behind them. Not just how many, but which pages, which formats and which topics get picked.
This is my second look at this data. My Microsoft Copilot citations study covers the growth curve and how Copilot searches, and my AI visibility case study covers what I changed on the site. This study is about something different: what kind of page gets cited, and whether it's the same page Google rewards.
The key findings at a glance
- My site earned 708,200 Bing AI citations in the 3-month view read on 30 October 2026.
- In the latest week I could read (23 to 27 October 2026), that was 142,700 citations in 5 days, or 28,540 a day.
- Comparison pages (X vs Y and alternatives) were 0 of the 60 most-cited pages, despite being 1,474 pages, or 6.9% of my site.
- Verb-led how-to pages (delete, access, pin, see, watch...) were 15.0% of the top 60 but only 1.7% of the site: 8.8 times their share.
- The top 60 Bing-cited pages earned 85 Google clicks in October 2026, which is 0.40% of my site's Google clicks.
- My single most-cited page had 113,000 Bing AI citations and 1 Google click in the same month.
- The top 1% of cited pages earned 38.6% of all citations, and the top 60 pages earned 53.9%.
- One topic (offline and no-internet games) earned 29.8% of all citations from just 2.0% of my pages.
- AI tools pages are 11.4% of my site but earned only 2.4% of the top 60's citations.
What gets cited by AI? The short answer from my data
On my site, pages that answer one specific, practical task get cited. In my top 60, that meant pages about doing something on a platform people already use: Facebook, Twitter/X, Instagram, WhatsApp, Gmail, plus free things to do offline.
Pages built to help someone choose between two products did not make the top 60 at all. Neither did templates, statistics pages or study pages.
One caution before the detail. Bing only shows a table of the 60 most-cited pages, so everything about page type, topic, age and Google overlap below is about those 60 pages. They're the top 4.35% of the 1,379 pages Bing had cited, and they hold 237,977 of 441,300 citations (53.9%) in the 21 July to 18 October 2026 view.
Which page types get cited by AI?
I sorted every page by its URL slug into a type (how-to, best-X list, X vs Y and so on), then compared each type's share of the top 60 with its share of all 21,474 live pages on my site.
| Page type | Pages in top 60 | Share of top 60 | Share of my site | Lift |
|---|---|---|---|---|
| Verb-led how-to (delete-, access-, pin-...) | 9 | 15.0% | 1.7% | 8.8x |
| Brand marketing-strategy analysis | 3 | 5.0% | 0.6% | 8.2x |
| Free-X pages | 7 | 11.7% | 4.5% | 2.6x |
| Question-led slugs (how many, why, where, can) | 12 | 20.0% | 7.7% | 2.6x |
| Reviews | 2 | 3.3% | 3.1% | 1.1x |
| How-to (how-to-, how-do-) | 8 | 13.3% | 14.4% | 0.9x |
| Best-X lists | 5 | 8.3% | 10.2% | 0.8x |
| Prompts | 1 | 1.7% | 2.1% | 0.8x |
| X vs Y | 0 | 0% | 3.0% | 0 |
| Alternatives | 0 | 0% | 3.9% | 0 |
| Templates | 0 | 0% | 0.8% | 0 |
| Statistics and studies | 0 | 0% | 0.3% | 0 |
Verb-led how-to pages were cited at 8.8 times their share of my site. These are slugs that start with the action, like deleting, accessing, pinning or watching something. Brand marketing-strategy breakdowns (Wix and Squarespace) came close at 8.2 times.
Plain "how-to" slugs sat at par: 13.3% of the top 60 against 14.4% of the site. So it's not the words "how to" that matter. It looks more like the pages that win are the ones tied to one concrete action.
Why are comparison pages missing?
Comparison pages earned zero places in the top 60. I've got 646 X vs Y pages and 828 alternatives pages, 1,474 in all. If the top 60 had been picked at random from my site, the chance of getting none of them is 1.4%.
I can't tell you why from this data. It could be the topics those pages cover, the kind of question Copilot gets, or the fact that most of them are newer. What I can say is that on my site, in this Bing report, comparison pages are not where the citations are.
Does one format earn more per page?
Not once you take out the star page. Excluding my #1 page, citations per page for the bigger groups ranged from 1,553 (other) to 3,154 (best-X lists), with how-to pages at 2,264 and verb-led how-tos at 1,817. The over-representation is mostly about how many pages of a kind made it in and which topics they sit on, not one format earning far more each.
Which topics get cited most?
| Topic | Pages in top 60 | Share of top 60 citations | Share of my site |
|---|---|---|---|
| Offline and no-internet games | 9 | 55.2% | 2.0% |
| 15 | 12.9% | 3.6% | |
| Twitter/X | 8 | 7.7% | 1.3% |
| 9 | 5.5% | 6.1% | |
| Product review sites | 2 | 4.7% | 0.1% |
| Wix and Squarespace marketing | 3 | 4.6% | 0.4% |
| Remote and Amazon jobs | 5 | 2.6% | 3.2% |
| AI tools (ChatGPT, Claude, Gemini) | 3 | 2.4% | 11.4% |
| 2 | 1.5% | 2.7% |
One topic, offline and no-internet games, earned 29.8% of all my Bing AI citations from 2.0% of my pages. Facebook pages were the next biggest group, with 15 of the top 60 pages.
The surprise for me was AI tools. They're the topic I write about most, 11.4% of my live pages, yet they earned only 2.4% of the top 60's citations. More pages on a topic did not mean more citations. If you want to see the star page, it's my no-internet games guide.
Do AI citations follow Google rankings?
Mostly not. I matched the top 60 Bing-cited pages against my Google Search Console data for 1 to 28 October 2026.
| Measure | Top 60 Bing-cited pages | All my live pages |
|---|---|---|
| Had any Google impressions | 57 of 60 (95%) | 53.7% |
| Got at least one Google click | 24 of 60 (40%) | 12.1% |
| Google clicks earned | 85 (0.40% of site) | 21,183 |
| Google impressions earned | 24,301 (0.75% of site) | 3,258,520 |
| Median average Google position | 8.8 | n/a |
The 60 pages that earned 53.9% of my Bing AI citations earned 0.40% of my Google clicks. Google does see them: 57 of the 60 had impressions and 40 of those 57 had an average position of 10 or better. But 36 of the 60 got no Google clicks at all.
My most-cited page had 113,000 Bing AI citations against 1 Google click, 41 impressions and an average position of 11.2. And of Google's top 60 pages by clicks, only 1 is also in Bing's top 60.
The rank correlation between Bing citations and Google clicks across the 60 pages was 0.27, which is weak. With Google impressions it was -0.04, which is none. Ranking on Google's first page is common among the cited pages, but more Google visibility does not go with more AI citations.
Are AI citations concentrated on a few pages?
| Slice of cited pages | Citations | Share of 441,300 |
|---|---|---|
| Top 1 page | 113,000 | 25.6% |
| Top 5 | 135,800 | 30.8% |
| Top 14 (the top 1%) | 170,300 | 38.6% |
| Top 20 | 186,900 | 42.4% |
| Top 60 | 237,977 | 53.9% |
| All other cited pages (about 1,319) | 203,323 | 46.1% |
The top 1% of my cited pages earned 38.6% of citations. That's concentrated, but it's loosening.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
My #1 page's share fell from 30.2% on 11 October to 29.4% on 14 October to 25.6% on 21 October 2026. Its own count kept rising, to 113,000. Its share fell because the rest of the site grew faster. Separately, between 30 July and 14 October the number of distinct pages Bing had cited went from 198 to 1,379, 7.0 times as many.
Does page age or length matter for AI citations?
Neither shows a link to how many citations a page gets once it's in the top 60. But two patterns stood out.
Brand-new pages rarely make the top. Only 3 of the 60 most-cited pages (5%) were 30 days old or less, while 65.8% of my live pages are that new. 42 of the 60 (70%) were 31 to 90 days old, against 15.2% of the site. The median age was 64 days.
The most-cited pages are longer than my typical recent page. By my estimate the top 60 have a median of about 1,970 words of body text, and 58 of the 60 have 1,500 or more. Across 6,878 recently fetched pages on my site the median is 1,180 words, and 23.6% reach 1,500. The two were measured slightly differently, so treat this as a direction, not a precise gap.
Answer-first openings are the norm, not the difference. 59 of the 60 most-cited pages open with a direct answer, but so do nearly all my pages: 95.6% of recent pages pass the same automatic test. So this data can't show that answer-first pages get cited more.
How fast are Bing AI citations growing?
| Read date (2026) | 3-month citations | Avg cited pages a day |
|---|---|---|
| 30 Jul | 23,400 | 40 |
| 1 Sep | 56,300 | 70 |
| 15 Sep | 84,500 | 113 |
| 11 Oct | 328,700 | 257 |
| 21 Oct | 441,300 | 344 |
| 25 Oct | 594,900 | 457 |
| 30 Oct | 708,200 | 574 |
The 3-month total reached 708,200 citations on the 30 October 2026 read. In the 7-day table for 23 to 27 October, Bing cited an average of 2,500 of my pages a day. My best single day on any read was 35,200 citations on 20 October.
Growth came from both sides: more pages being cited, and each one being cited more often. Searches in my top 60 queries were short, with a median of 4 words, and only 3 of those 60 queries were phrased as a question.
How to get cited by AI: what this means for your business
This is one website and one AI engine, so read these as what I'd do next, not rules. With that said, here's how I'd act on it.
- Write for one task per page. The pages that over-performed were tied to one concrete action. Pick the jobs your customers search for and give each its own page that does it from start to finish.
- Don't assume your Google winners are your AI winners. Check your AI citations separately. On my site the two lists barely overlap.
- Don't judge a topic by how much you've written about it. My biggest topic by page count earned a small share of citations. Look at where citations land and build around that.
- Give new pages time. Few of my top pages were under a month old. A page that isn't cited in week two may still get there.
- Cover the subject in full. The most-cited pages were longer than my typical page. Answer the question first, then cover the follow-up questions too.
- Track it. If you have Bing Webmaster Tools, the AI Performance report shows your citations. If you don't, start with a quick check of where you stand.
Methods and limitations
Data sources. Citation figures come from my Bing Webmaster Tools AI Performance report, read on several dates between 30 July and 30 October 2026. A citation is Bing, Copilot or a partner linking one of my pages as a source in an AI answer. The page table is from the 21 October 2026 read (3-month view, 21 July to 18 October). The 708,200 total is from the 30 October read, which shows totals only. Google figures come from my Search Console, page level, 1 to 28 October 2026.
Classification. Each page got one type and one topic from its URL slug, using fixed rules applied in order. The live baseline is all 21,474 live pages on my site as of 30 October 2026. Page age uses the WordPress publish date. Word counts come from fetching the 60 live pages on 31 October 2026. Correlations are Spearman rank correlations.
Limitations.
- One website and one engine (Bing, Microsoft Copilot and partners). This is not ChatGPT, Perplexity or Google AI Overviews data.
- Page-level findings cover only the top 60 cited pages. The other pages, about 46.1% of citations, can't be broken down from Bing's report.
- Bing rounds page values to 3 significant figures, and 3-month windows shift between reads.
- The Google window (1 to 28 October) differs from the Bing page window (21 July to 18 October).
- One page holds 25.6% of citations and one topic holds 29.8%, so any finding that includes them is dominated by them.
- Slug-based classification is rough and the sample is 60 pages. Treat the results as indicative.
- Everything here is correlation. Nothing in this data proves what causes a page to be cited.
Frequently asked questions
What gets cited by AI search?
On my site, in Bing's AI Performance report, the most-cited pages answered one practical task, often on a platform like Facebook, Twitter/X or Instagram. Verb-led how-to pages were cited at 8.8 times their share of the site, and comparison pages were absent from the top 60.
Do AI citations follow Google rankings?
Not on my site. The 60 pages with the most Bing AI citations earned 85 Google clicks in October 2026, 0.40% of my total. The rank correlation with Google clicks was a weak 0.27, and with impressions it was -0.04.
Are comparison pages cited by AI?
Not in my top 60. I've got 1,474 X vs Y and alternatives pages, 6.9% of my site, and none were among the 60 most-cited pages in the 21 July to 18 October 2026 view.
How many Bing AI citations did your site get?
708,200 in the 3-month view read on 30 October 2026. In the week of 23 to 27 October it averaged 28,540 citations a day.
Where can I see my own AI citations?
Bing Webmaster Tools has an AI Performance report showing citations from Bing, Copilot and partners. It shows totals, daily figures, top pages and the searches behind them.
Want to know if AI cites your business?
Run your site through my free AI Visibility Checker for a quick first read on how visible you are to AI tools. If you'd like help getting cited, my AI SEO services are where I do this work for business owners, using the same approach I've tested on my own site.
