When Students Ask AI About University Rankings: How Citable Are Universities' Own Ranking Pages? A Webometrics Study

Prospective students increasingly ask AI assistants and conversational search tools direct questions about a university's ranking standing rather than reading a ranking table themselves. For an AI system to answer accurately and attribute its answer correctly, the university's own page describing its ranking achievements needs to be retrievable, dated, and readable as real text rather than locked inside an image or a script-rendered widget. Webometrics examined the official "rankings" or "achievements" pages of a purposive sample of 15 universities across seven world regions, collected on 26 September 2026. A qualifying official page was found and inspected at 13 of the 15 institutions. Among those 13, 8 (61.5%) stated a specific ranking year or edition, only 4 (30.8%) linked out to the external ranking organization that produced the figures, and 8 (61.5%) presented the ranking information as plain, crawlable HTML text. One institution's ranking page consisted of a single image with no crawlable ranking text at all, and one presented its figures inside a JavaScript-rendered dashboard with no ranking data in the static page source. Across the 15 domains studied, 11 robots.txt files were retrievable; one explicitly disallowed two documented AI/data-scraping crawlers site-wide. The pattern suggests that many universities' own ranking pages are not yet built with AI and search retrieval in mind, independent of how well those universities perform in any ranking.

Key Findings

• A qualifying official university page presenting ranking standings was found and inspected at 13 of the 15 sampled institutions (86.7%); no such page could be located for the remaining 2 after a reasonable search.
• Of the 13 inspected pages, 8 (61.5%) stated a specific year or edition for the ranking figures shown; 3 (23.1%) stated no year or edition at all.
• Only 4 of the 13 inspected pages (30.8%) contained a hyperlink back to the external ranking organization's own website as the source of the figures.
• Ranking information appeared as plain, crawlable HTML text at 8 of 13 pages (61.5%); at 1 page (7.7%) it existed only inside an image file, and at 1 page (7.7%) it was not present in the static page at all, only rendered client-side by JavaScript.
• 11 of the 13 pages (84.6%) displayed some visible date or copyright signal, but 4 of those 11 (36.4%) showed a date inconsistent with, or older than, the ranking data shown on the same page.
• Schema.org structured data could not be verified as present or absent for any of the 13 pages, because this study's retrieval method could not access the underlying HTML source needed to detect it a disclosed methodological limitation, not a finding.
• Of 11 retrievable robots.txt files across the 15 domains, 1 (University of Cape Town) explicitly disallowed two documented AI/data-scraping crawlers site-wide; the other 10 contained no directive naming any of the four AI crawlers checked.

Why This Matters for Universities

Students, journalists, and prospective applicants are no longer only reading ranking tables directly; a growing share of that discovery now happens through conversational search and AI assistants that answer a question first and cite a source second. When a university wants its own account of its ranking standing to be part of that answer rather than an outdated or third-party summary the technical form of its own ranking page matters as much as the ranking result itself. A page that never states which year's figures it is showing, that keeps its numbers inside an image, or that renders them only after a script runs in a browser, is harder for any automated system, AI or otherwise, to read, date, and attribute correctly. This is a question of retrievability and clarity, not of institutional prestige: a page's technical form is unrelated to how well an institution performs in any ranking.

Authoritative Background

Search engines and AI systems generally rely on two related capabilities: a crawler that can fetch and parse a page's content, and a set of signals visible dates, source links, and machine-readable markup that help a system understand what the content means and when it was produced. Google Search Central's crawling and indexing documentation explains that content must be present in a form Google's crawler can access and parse in order to be eligible for indexing and for AI-powered search features; content that is generated only by client-side JavaScript after the initial page load can be missed or delayed if a site is not built with server-side or pre-rendered content for those elements. The Internet Engineering Task Force's RFC 9309, which formally standardizes the Robots Exclusion Protocol, defines how a robots.txt file's directives are structured and interpreted by any crawler that chooses to honor them, including the AI-specific crawlers operated by OpenAI (documented as GPTBot), Anthropic (documented as ClaudeBot), and Google (documented as Google-Extended). Schema.org provides the vocabulary including the Article, Dataset, and WebPage types that a page can use to state machine-readably what it is and when it was last modified, which is one of the signals Google Search Central's own structured-data documentation associates with better content understanding. None of these sources describe a specific AI system's citation-selection process beyond what each company documents publicly, and this study does not claim otherwise.

Research Question

When a university presents its own ranking or achievement standing on an official web page, does that page carry the features associated with reliable retrieval and attribution by search engines and AI systems: a stated ranking year or edition, a link to the external source of the figures, ranking information present as real crawlable text rather than only an image, a visible freshness signal, and technical crawler access via robots.txt?

Methodology

Collection date: 26 September 2026, in a single collection session.

Population: Official web pages, hosted on a university's own domain or an official institutional subdomain, that present the institution's standing in one or more international university rankings.

Sample: 15 universities, purposively selected to represent seven world regions and drawn from institutions widely recognized for research activity and international visibility: 2 from North America, 2 from Europe, 4 from Asia, 1 from Oceania, 2 from Africa, 2 from Latin America, and 2 from the Middle East.

Selection method: Judgment sampling, not random selection. Institutions were chosen for regional spread and for being large, internationally oriented research universities likely to maintain an official ranking or achievements page, rather than through a random draw from a defined population of all universities. This is a deliberate limitation, disclosed below.

Inclusion rule: An institution was treated as having a qualifying page only if an official page or news item, hosted on the institution's own domain, was found presenting specific ranking standings. Pages that discussed university facts generally without mentioning rankings were not treated as qualifying substitutes.

Variables recorded for each qualifying page: whether a specific ranking year or edition was stated; whether the page linked to the external ranking organization's own website; whether the ranking information appeared as selectable HTML text, as an image or graphic, as a PDF, or as a mix; whether any visible last-updated, "as of," or copyright date signal was present, and whether it was consistent with the ranking data shown; whether schema.org structured data could be detected; and, separately for each domain, whether robots.txt was retrievable and whether it named any of four documented AI crawlers (GPTBot, ClaudeBot, Google-Extended, PerplexityBot).

Evidence sources: Direct retrieval of each institution's official page and, separately, each domain's robots.txt file.

Analysis approach: Descriptive tabulation of the recorded variables across the determinate sample, with exact counts and denominators reported throughout.

Limitations: See the Limitations section below.

Results

Official ranking pages (13 of 15 institutions with a qualifying page)

Harvard University and the University of Tokyo are not included in this table: after a reasonable search, no official page presenting specific ranking standings could be located on either institution's domain.

Across the 13 determinate institutions: 8 (61.5%) stated a specific ranking year or edition, 1 (7.7%) gave partial or inconsistent edition information, 1 (7.7%) was ambiguous, and 3 (23.1%) stated none. Only 4 of 13 (30.8%) ETH Zurich, the University of Melbourne, Universidade de São Paulo, and UNAM linked directly to the external ranking organization's own website; the remaining 9 (69.2%) named or displayed ranking results without a source hyperlink. Ranking information was present as plain, crawlable HTML text at 8 of 13 pages (61.5%); mixed with supporting images at 3 of 13 (23.1%); confined entirely to a single image with no crawlable text at 1 of 13 (7.7%, American University of Beirut); and absent from the static page altogether, rendered only by client-side JavaScript, at 1 of 13 (7.7%, University of Toronto).

Eleven of the 13 pages (84.6%) displayed some visible date or copyright signal; the remaining 2 (University of Nairobi and American University of Beirut) had none. Of the 11 with a visible signal, 4 (36.4%) Tsinghua University, Universitas Indonesia, Universidade de São Paulo, and UNAM showed a date that was either frozen well in the past, older than the ranking data displayed on the same page, or an internally inconsistent forward-dated range.

Schema.org structured data (JSON-LD or microdata) could not be confirmed as present or absent for any of the 13 pages. This study's retrieval method surfaced rendered page text rather than raw HTML source, so structured-data presence is reported as not verifiable rather than as an observed absence; this is stated explicitly as a limitation rather than a finding.

Robots.txt and documented AI crawlers (15 domains)

Robots.txt was retrievable for 11 of the 15 domains studied. It could not be retrieved for 4 domains, each for a distinct, disclosed reason: two domains (ETH Zurich, Tsinghua University) returned a "not found" response for every robots.txt path attempted; one domain (Universitas Indonesia) returned an access-management block page in place of robots.txt content; and one domain (University of Nairobi) could not be reached due to a certificate-validation failure during the connection itself. These are treated as genuine technical observations about the domains in question, not as evidence about crawler policy, since no robots.txt content could actually be read.

Among the 11 retrievable robots.txt files, 10 (90.9%) contained no directive naming any of the four documented AI crawlers checked (GPTBot, ClaudeBot, Google-Extended, or PerplexityBot); at these 10 domains, such crawlers would fall under each site's general, and in every one of these 10 cases permissive, wildcard rule. One domain, the University of Cape Town, explicitly named and fully disallowed two documented AI/data-scraping crawlers, GPTBot and Bytespider, site-wide. None of the 11 retrievable robots.txt files named ClaudeBot, Google-Extended, or PerplexityBot specifically, in either direction.

Analysis

Two patterns stand out. First, stating a ranking's edition year and presenting it as real text are the two features most universities in this sample already get right, while linking back to the external source and keeping the date signal internally consistent with the data shown are the two features most commonly missing. A page that reports a global ranking position without a year attached, or without any link back to where that figure came from, gives an automated reader little basis for confirming when the claim was true or where it originated a limitation for a human reader making the same judgment. Second, the sample contained two clear-cut technical barriers to any form of retrieval, human or automated: one institution's ranking figures existed only as pixels inside an image file, and another's existed only inside a script that had to execute before any ranking data appeared in the page at all. Both cases mean that, at the moment this sample was collected, a system that reads static page content would find no ranking text to work with on those two specific pages, independent of anything the underlying institution's actual ranking performance shows.

The single explicit AI-crawler block observed, at the University of Cape Town, is worth noting as a distinct case from the rest of the sample: it reflects a deliberate site-wide policy choice rather than an accidental gap, and it is consistent with the range of AI-crawler postures Webometrics has documented in its earlier study of AI-crawler access in university robots.txt files, where blocking policies vary considerably between institutions even within the same sample. It cannot be inferred from this single observation whether the policy was applied with the ranking page specifically in mind, since the disallow rule covers the entire site rather than that one page.

University Examples

The University of Toronto's ranking summary is maintained on an official data-and-reporting subdomain, but the specific figures load through an interactive, JavaScript-rendered dashboard; the underlying static page carries only a title and navigation content, with no ranking names, positions, or figures present as text before any script executes. The American University of Beirut's official rankings page took the opposite technical form: it is built around a single image file containing the ranking summary, with no separate text version of the same figures on the page. The University of Cape Town's page, by contrast, presented most of its ranking content as ordinary HTML text with a stated current-year framing, but the same domain's robots.txt separately disallowed two documented AI/data-scraping crawlers across the entire site. These examples describe measurable technical characteristics observed on the dates and pages specified; they are not evaluations of the underlying institutions' academic standing or overall digital strategy.

Implications for Higher Education

For university communications, digital strategy, and web-management teams, the practical implication is that a ranking or achievements page can be strengthened for retrieval and attribution without any change to the ranking result itself: stating the edition year for every figure shown, linking each figure to the ranking organization's own published results, ensuring the ranking text exists as real HTML rather than only an image or a script-rendered widget, and keeping a genuine, consistent "last updated" signal are all changes within a communications or web team's direct control. None of these changes affects how any ranking body calculates its results, and none of them is a guarantee of improved search visibility or AI citation; they simply remove technical barriers to a system reading and dating the content that is already there. Readers interested in the broader set of signals that make an institutional website well understood by search and web systems can also see Webometrics' study of digital presence signals that matter.

Recommended Actions

• State the specific year or edition alongside every ranking figure published on an official page, rather than presenting a bare position or score.
• Link each cited ranking figure to the external ranking organization's own published results page, so a reader or an automated system can verify the source directly.
• Ensure ranking figures exist as selectable HTML text on the page itself, in addition to any supporting graphic, rather than relying on an image or a script-rendered widget as the only source of the information.
• Maintain a genuine, visible "last updated" or "as of" date on the page, and keep any copyright-year notice consistent with the data actually shown, rather than leaving a stale template value in place.
• Periodically retrieve the domain's own robots.txt file directly to confirm it reflects the institution's actual, current intent for AI-crawler access, rather than assuming a default configuration is still in effect.

Limitations

This study used a purposive, judgment-based sample of 15 large, internationally oriented research universities rather than a random or stratified probability sample of all universities; the findings describe this specific sample on this specific date and should not be generalized to universities generally. Schema.org structured-data presence could not be verified for any of the 13 pages because this study's retrieval method exposed rendered page text rather than raw HTML source; this is reported as a methodological limitation rather than as evidence that structured data is absent. Robots.txt could not be retrieved for 4 of the 15 domains, for reasons ranging from a "not found" response to an access-management block to a certificate-validation failure, so no crawler-access conclusion is drawn for those domains. Websites change frequently; a page's technical form on 26 September 2026 may not reflect its form at another time. Finally, this study observes association between page characteristics and retrievability-relevant features; it does not measure whether any specific AI system did or did not cite any of these pages, and it does not establish that any particular technical change would cause a change in citation or ranking outcomes.

Conclusion

Webometrics examined the official ranking pages of 15 universities across seven world regions on 26 September 2026 and found a qualifying page at 13 of them. Among those 13, most stated which year's ranking they were describing and presented the information as ordinary text, but fewer than a third linked back to the external source of the figures, and more than a third of the pages with a visible date signal showed a date inconsistent with the ranking data displayed. Two pages in the sample kept their ranking figures outside the reach of a static-text reader entirely, one behind an image and one behind client-side rendering. These are technical, addressable characteristics of how a ranking claim is published, separate from the ranking result itself, and they describe a pattern worth revisiting as universities' own pages become one more surface that search engines and AI assistants read, date, and attempt to attribute. Readers can review current standings on the Webometrics ranking results and the Webometrics methodology page.

Sources

• Google Search Central, crawling, indexing, and JavaScript SEO documentation developers.google.com/search
• Google Search Central, structured data general guidelines developers.google.com/search/docs/appearance/structured-data
• Internet Engineering Task Force, RFC 9309, "Robots Exclusion Protocol"
• OpenAI, GPTBot documentation
• Anthropic, ClaudeBot documentation
• Google Search Central, Google-Extended documentation
• Schema.org, Article and WebPage vocabulary reference
• Direct retrieval of official university pages and robots.txt files, collected 26 September 2026 (Webometrics original research)

Suggested Citation

Webometrics by UNIRANKS. (2026). When Students Ask AI About University Rankings: How Citable Are Universities' Own Ranking Pages? A Webometrics Study. Webometrics.org. https://www.webometrics.org/blog/ai-search-visibility/university-ranking-page-ai-citability-study