Home › Resources › Open data

What Microsoft Copilot cites in fashion ecommerce: 2,700 AI citations analysed

Over three months, Microsoft Copilot and its partners cited renderedfits.com about 2,700 times. 71% of the sampled citations answered research questions, and 1.3% answered a buying question. This page publishes what the AI looked up, which pages it chose and how that changed from the three months before, using the site's own Bing Webmaster Tools data. It is one fashion technology website, not the whole market, so read it as a detailed case study.

Where these figures come from: our own research. The sample, method and limits are set out in How to quote these figures. How we choose and check sources: editorial policy.

By · Published 20 November 2026 · Data for 18 May to 17 November 2026, exported 17 and 18 November 2026

Key findings

  1. About 2,700 citations of renderedfits.com by Microsoft Copilot and its partners between 18 August and 17 November 2026, on an average of 6 cited pages a day.
  2. 3.0 times as many page citations as the three months before: 2,368 citations on 40 pages, up from 779 on 29 pages (18 May to 18 August 2026).
  3. 71% of the 701 citations in Bing's sample of grounding queries came from queries labelled Research (497). Informational queries gave 12% (85).
  4. 1.3% of sampled citations came from a buying query (9 citations, one query labelled Commercial). No query labelled Comparison cited the site in either period.
  5. 45% of page citations went to research and statistics pages, 42% to long guides, 7% to comparison and pricing pages, 4% to size conversion tools and 2% to the homepage and brand pages.
  6. 44% of page citations went to just three pages, and 78% to the top ten.
  7. 2 of 27 sampled grounding queries also appeared among the 11 for the previous three months. The questions AI assistants look up changed almost completely.
  8. 6.4 words was the average length of a grounding query, written like keywords. The 793 Bing searches that showed the site in the same period averaged 8.8 words, and 71% had six or more.
  9. 23.7% was the median citation share on the sampled queries, from 2.7% to 58.2%.
  10. 46 clicks came from Bing web search in the same three months, from 2,100 impressions. A citation shows a site as a source in an AI answer; it is not a visit.

What did Copilot look up?

A grounding query is the search an AI assistant runs to find sources before it writes its answer. These are the 27 grounding queries Bing listed for the site, with Bing's own intent label, the number of citations and the site's citation share (the site's part of all citations Bing recorded for that query). Spelling is as Bing shows it.

By our grouping, GEO and AI search questions produced 170 of the 701 sampled citations (24%), virtual try-on and fitting rooms 164 (23%), returns 142 (20%), AI personalisation and shopping agents 119 (17%), purchase confidence 86 (12%) and conversion rates 20 (3%).

Grounding queries citing renderedfits.com, 18 August to 17 November 2026. Source: Bing Webmaster Tools, AI Performance.
Grounding queryBing intentCitationsCitation share
GEO strategies fashion e-commerceResearch9017.89%
percentage of fashion returns due to fit or appearanceInformational8231.06%
AI personalization future fashion trends online shoppingResearch7422.16%
average return rate online fashion industry benchmarkResearch5032.26%
GEO strategy fashion brand global expansionResearch4428.39%
retailers live video try-on clothingOthers3958.21%
GEO Global Expansion Strategy fashion brandResearch3628.57%
strategies reduce hesitation high-ticket ecommerce purchasesResearch3521.60%
impact of ai shopping agents on fashion ecommerceResearch3327.73%
ai virtual try-on clothes technologyOthers2619.40%
factors influencing purchase confidence Ultimate ProductsResearch2517.36%
virtual fitting room technology try on clothesOthers2218.80%
fashion brands virtual try-on technology clothesResearch1838.30%
virtual fitting roomOthers1823.68%
ultimate products purchase confidence factorsResearch1625.81%
uk fashion ecommerce checkout conversion rate average basket £45Research1541.67%
future of virtual try-on fashion ecommerceResearch1115.28%
clothing brands virtual try-on technologyResearch1155.00%
virtual fitting technology reduce returns improve customer satisfaction retail softwareResearch106.41%
virtual try-on software pricing models costResearch915.79%
best shopper fit agent apps Shopify apparel USCommercial922.50%
factors influencing purchase confidence ultimate products companyResearch722.58%
virtual try-on impact conversion rates fashion ecommerceResearch723.33%
european luxury coat size conversion us ukUtility52.67%
best strategies to reduce hesitation for high ticket ecommerce purchasesResearch337.50%
ai sizing apps shopify pricingResearch330.00%
generative virtual try-on technologyInformational337.50%

Download: grounding-queries.csv (with the pages cited for each query).

Which pages did Copilot cite?

The fifteen most cited pages, with citations in the latest three months and the three months before. Pages that are original data or collect published statistics took the largest share. Product pages for the app itself were rarely cited, and the homepage fell from 90 citations to 23 while citations of the whole site tripled.

Download: cited-pages.csv (all 52 pages cited in either period).

What does this mean for fashion brands?

This is one site, so treat these as observations to test on your own data rather than rules. Four things stood out.

AI assistants cite the research stage, not the shop window. Almost every citation answered a question a buyer asks before they are ready to choose: what return rates are normal, how the technology works, what other brands do. A brand that only publishes product pages has little for an assistant to quote at that stage.

Pages with dated, linked figures were cited most. The most cited data page, the Fashion Returns Index, gives each figure a named source and a date. Assistants answering a "percentage of" or "benchmark" question need a number they can attribute.

The questions move fast. Only 2 of the 27 sampled queries appeared in both periods, so chasing single phrasings is a weak plan. Covering a topic thoroughly held up better: each of the top three pages was cited for more than one query.

Product facts are a separate problem. When an assistant does reach the buying stage, it needs facts about the product itself, such as sizes and measurements. Our AI sizing study found that for 80% of 1,383 Shopify clothing stores the sampled product reached AI shopping assistants with no size chart. See our GEO guide for fashion and the AI agent fit check.

How can you see Copilot citations for your own site?

Verify your site in Bing Webmaster Tools, then open AI Performance. It shows total citations from Microsoft Copilot and its partners, the pages cited, and the grounding queries behind them, with Bing's intent and topic labels. Pick a date range to compare periods; there is no side-by-side view, so export each period separately.

Bing's query table is a sample: for this site it listed 27 queries holding 701 of about 2,700 citations. Clicking a query filters the pages view to the pages cited for it.

Method and limits

All figures come from Bing Webmaster Tools for www.renderedfits.com, exported by Rendered Fits on 17 and 18 November 2026. "About 2,700" is Bing's chart total for 18 August to 17 November 2026; the page table for the same period sums to 2,368 across 40 pages, and the query table to 701 across 27 queries. Intent and topic labels are Bing's, generated by AI. The page types (research and statistics, guides, comparison and pricing, size tools, homepage and brand) and the topic groups are our own grouping of the URLs and queries; the CSV files let you regroup them.

The data describes one site that publishes a lot of fashion research, so its mix of pages shapes its mix of citations. Bing does not say which assistant made each citation or how many people saw the answer. Search clicks and impressions are Bing web search figures for the same three months.

How to quote these figures

Name the source, the site and the period, and link to this page. For example:

"71% of 701 sampled Copilot citations of a fashion ecommerce site answered research questions, and 1.3% a buying question." Rendered Fits, What Microsoft Copilot cites in fashion ecommerce (Bing Webmaster Tools data, 18 August to 17 November 2026).

The figures and the two CSV files are free to reuse under CC BY 4.0. Every statistic is also in statistics.md, and our other studies are on the open data page and in clothing size and fit statistics 2026.

Questions

What does Microsoft Copilot cite for fashion ecommerce questions?

In our data, mostly research. Of 701 citations in Bing's sample of grounding queries for renderedfits.com (18 August to 17 November 2026), 71% came from queries Bing labels Research, such as return rate benchmarks and GEO strategy. One buying query produced 9 citations (1.3%), and no comparison query cited the site.

Which kinds of page does Copilot cite most?

Research and statistics pages received 45% of the 2,368 page citations Bing listed, and long guides 42%. Comparison and pricing pages received 7%, size conversion tools 4%, and the homepage and brand pages 2%. Three pages took 44% of all citations.

What is a grounding query?

It is the search an AI assistant runs to find sources before it writes an answer. Bing Webmaster Tools shows the grounding queries that led Microsoft Copilot and its partners to cite a site, with the number of citations and the site's citation share.

Do AI citations bring visitors?

Not many on their own. In the same three months the site was cited about 2,700 times by Copilot and its partners, while Bing web search sent 46 clicks. A citation shows the site as a source in an answer; it is not a visit.

How can I see Copilot citations for my own site?

Verify the site in Bing Webmaster Tools and open AI Performance. It shows total citations, cited pages and grounding queries for Microsoft Copilot and its partners, with a date range picker.

Can I use these figures?

Yes. The figures and the two CSV files are our own data, free to reuse under CC BY 4.0 with credit to Rendered Fits and a link to this page.

About Rendered Fits. Rendered Fits is a Shopify app for fashion brands that recommends a size for supported products from the brand's own size chart, and can show shoppers an optional AI try-on preview. There is a Free plan (one product, 5 try-ons a month, with a PREVIEW watermark). Complete is $349 a month plus any applicable taxes, with a 14-day trial of 50 try-ons in total, then 1,000 try-ons per paid cycle. Extra try-ons cost $0.50 each, but only if you turn them on and set a spending cap. Enterprise is $999 a month plus any applicable taxes, with 5,000 try-ons per paid cycle, then $0.35 each extra on the same terms, and no trial.

See also: GEO for fashion · AI shopping agents statistics · Fashion Returns Index · open data · AI sizing study

Free plan availableBilled through Shopify

Install