Which URLs are cited most often?
Use Bing AI Performance's Pages view to identify cited URLs and compare the distribution across destination guides, sea-temperature pages and explanatory articles.
First-party case study · Bing AI Performance · SnorkelForecast
10,817 Bing AI citations. Here's what the data revealed.
SnorkelForecast's citation appearances grew 23.6% from September to October 2026. But the surprising part was elsewhere: fewer pages were cited per day while total citations increased. That gives us a much better question to investigate than the headline number alone.
The central finding · September vs October 2026
More citations. Fewer cited pages per day.
Bing's daily export records both the number of cited appearances and the number of unique pages cited that day. Looking at both changes how we interpret the trend.
+23.6%Monthly citations3,278 → 4,052
−11.0%Average cited pages/day28.1 → 25.0
+34.5%Citation intensity3.89 → 5.23 citations per cited-page-day
| Metric | September | October | Change |
|---|---|---|---|
| Bing AI citations | 3,278 | 4,052 | +23.6% |
| Cited-page-days (sum of daily cited-page counts) | 843 | 775 | −8.1% |
| Average cited pages per day | 28.1 | 25.0 | −11.0% |
| Citations ÷ cited-page-days | 3.89 | 5.23 | +34.5% |
“Cited-page-days” is the sum of Bing's daily unique cited-page counts, not the number of unique URLs for the entire month. The ratio is our derived measure of citation intensity; it is not a per-URL citation count or evidence of which URLs were responsible.
The research question
Were a smaller set of source pages being cited more frequently?
That's one plausible explanation worth testing, not something the overview alone proves. Page-level citation and grounding-query reports would tell us what to investigate next.
181-day view
The growth trend across seven calendar months.
These are monthly sums of daily Bing AI citation counts from 9 May through 5 November 2026, including days with zero citations.
May 9–31
13
June
205
July
746
August
1,637
September
3,278
October
4,052
Nov 1–5
886
May and November are partial months. The chart compares absolute counts, not daily rates; October is the scale maximum.
10,817Citations in the full period9 May–5 November 2026
181Daily observationsExport includes days with zero
4,052October's cited appearances130.7 per calendar day
The information behind the site
An example of a source page designed to answer a real question.
SnorkelForecast publishes month-specific sea temperature data, comparisons and source notes. That is a concrete example of information design, even though these exports do not establish whether Bing cited this specific page.
Example traveller question
“How warm is the sea in Sardinia in May?”
Useful response: the historical baseline average is 18.2°C in May, compared with 26.1°C in August, for 2022–2025 observations. The comparison gives the reader seasonal context instead of a vague “warm in summer” answer.
The visible source
Sardinia sea-temperature page →
Published site content distinguishes historical monthly averages from recent forecasts and names the underlying Open-Meteo Marine / Météo-France data methodology.
This example is not a measured Bing grounding query or proof that the linked URL was cited. The page-level Bing data would be needed to make that claim.
The next investigation
Three questions that can produce genuinely new information.
Monthly citation totals describe the trend. Page and grounding-query evidence would help explain which source topics and content formats deserve attention.
What topics lead to those citations?
Link Bing's grouped grounding phrases to the cited-page view. These are retrieval-topic summaries, not the original user's questions.
What makes a citation-worthy answer?
Compare real cited pages for specific facts, original data, provenance, freshness and answer structure. Apply improvements and evaluate their effects with dated follow-ups.
The uploaded evidence includes the Bing overview trend and broad topic portfolio, not the detailed Pages or Grounding Queries export. We therefore don't present a fabricated URL-level citation ranking. See Bing's AI Performance documentation for how the views relate.
Separate measurement · Google web search
Search visibility is not the same as AI citations.
The accompanying Google Search Console workbook totals 44,337 Web search impressions from 18 May to 7 November 2026. Its Filters sheet specifies Web (text-based), not an AI Overview-only filter.
The latest 28-day period recorded 18,076 impressions, compared with 15,986 in the preceding 28 days, an increase of 13.1%.
Among the leading Google Web URLs were Sardinia sea temperature (4,098 impressions), Maui (2,758), and Best time to snorkel (1,655). These figures are not a ranking of URLs cited in Bing's AI experiences.
Transparent methodology
How these figures were calculated.
181 dated records
First-party overview export captured 9 November 2026, with Date, Citations and Cited Pages columns. Dates range from 9 May through 5 November 2026. Monthly citation totals are sums of each daily row, including zero-citation days.
How the ratio works
September citations ÷ 843 cited-page-days = 3.89. October citations ÷ 775 cited-page-days = 5.23. The change in this derived ratio is +34.5%; it is a summary indicator, not a claim that each page received that many citations.
Web-search context
Separate workbook export captured 9 November 2026, including Chart, Pages and Filters sheets. Chart dates cover 18 May to 7 November. The Web (text-based) filter does not establish AI-only impressions.
What the current evidence can't answer yet
- Which individual SnorkelForecast URLs drove Bing citation growth: the overview does not contain page-level rows.
- Whether the same pages were repeatedly cited across months: summing daily cited pages does not deduplicate URLs across days.
- What caused the change: we would need a dated implementation log, comparable periods and consideration of seasonality and Bing reporting changes.
- How many customers visited or converted from an AI answer: citation appearances are not clicks, referrals or conversions.
Bing describes this report as aggregated and sampled visibility data. The public page shares calculated findings and methodology, not private raw account exports.
From evidence to implementation
This is how I connect the parts.
SnorkelForecast
Observe cited appearances and find precise questions the existing reports leave unanswered.
RefreshAgent
The Search Console and analytics product I built to turn measurement into repeatable investigations.
SEO Kaizen Loop
Queue improvements, record baselines and revisit the same pages after changes.
Need an AEO consultant who can implement the work?
I work on source content, technical search foundations and measurement. We agree the investigation and delivery scope before starting.