Are Indonesian Universities Ready for AI Search? A Webometrics Study

Executive Summary

Google has stated publicly that structured data helps it understand page content and can make results "more engaging," and that its AI Overviews and AI Mode features draw on the same underlying Search index rather than requiring separate optimization. This raises a direct question for any national university system: are institutional websites built in a way that lets search and AI systems read them cleanly in the first place? Webometrics examined the public homepages and robots.txt files of ten prominent Indonesian universities, drawn from a purposive sample of nationally leading public and private institutions, in September 2026. Two of the ten sites could not be retrieved for review at the time of collection. Of the eight pages that could be reviewed, six had a working sitemap declared in robots.txt, seven carried a canonical tag, and only two showed any visible schema.org structured data, with the specific type of markup unconfirmed in both cases. None of the eight showed a technically correct hreflang implementation; Indonesian/English language variants were instead signaled through Open Graph locale tags or plain navigation links. One university's robots.txt pointed its sitemap directive at a non-functional placeholder domain. These are foundational, easily addressed technical gaps rather than judgments about institutional quality, and they matter because Google's own guidance ties AI-feature eligibility to ordinary Search indexability rather than any AI-specific markup.

Key Findings

1. 2 of 10 sampled university websites (20%) could not be retrieved for review at the homepage, the robots.txt endpoint, or both, preventing technical observation.
2. 6 of 8 reviewable sites (75%) declared a working XML sitemap in robots.txt; one declared no sitemap, and one declared a sitemap URL pointing to a non-functional placeholder domain rather than the university's own domain.
3. 7 of 8 reviewable sites (88%) carried a <link rel="canonical"> tag on the homepage; one did not show one in the content reviewed.
4. Only 2 of 8 reviewable sites (25%) showed any visible schema.org structured data (JSON-LD) on the homepage, and in neither case could the specific schema type be confirmed from the content reviewed.
5. 0 of 8 reviewable sites showed a confirmed <link rel="alternate" hreflang="..."> tag. Indonesian/English bilingual navigation was present on most sites, but implemented through Open Graph og:locale meta tags or manual /en/ navigation links rather than the HTML mechanism search engines use to serve the correct language variant.
6. Robots.txt disallow rules were narrow and conventional across the sample, blocking admin panels, login pages, uploads, or PDFs, with no evidence of blanket blocking of content sections.
7. In Webometrics' own analytics for the 90 days to 17 September 2026, Indonesia was the third-largest source of active users to Webometrics.org (7,640 users) and the highest-engagement large market (49.7% engagement rate; 1 minute 20 seconds average engagement time per user), indicating strong existing reader interest in Webometrics research from Indonesia, separate from and unrelated to the university-website findings above.

Why This Matters for Universities

A university's visibility in ranking systems, search engines, and AI-driven discovery depends in part on how accessible its research and institutional information are to the systems that surface them. Search engines and AI answer systems do not read a page the way a human does: they rely on crawlable HTML, canonical signals to avoid duplicate-content confusion, sitemaps to discover pages efficiently, and, where present, structured data to interpret entities such as courses, people, or organizations unambiguously. A site that a crawler cannot reach, or that gives conflicting or absent signals about its canonical URL and language variants, is harder to index correctly regardless of the quality of the content itself. For a national system with as large and linguistically bilingual a higher-education sector as Indonesia's, these are structural rather than content questions, and they are inexpensive to address once identified.

What the Authoritative Source Says

Google's developer documentation states that structured data "helps Google understand the content of the page," which "enables the display of search results, and helps us understand information across the web," and that pages with well-implemented markup have shown measurable gains in independently documented cases. Google requires that a schema type's required properties all be present for a page to be eligible for the corresponding rich result, and continues to recommend JSON-LD as the preferred format, tested against the Rich Results Test.

On AI features specifically, Google's documentation states: "there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary" beyond ordinary Search eligibility. A page must be crawlable, indexed, and eligible for standard Search snippets. Google's recommended practices are foundational: allow crawling via robots.txt and server configuration, maintain clear internal linking, provide content in readable text form, and keep structured data consistent with the visible page content. Google adds that publishers do not need to create special AI-specific files or markup, as the existing Search-indexing pipeline is what AI features draw from.

Read together, these two points mean that the technical baseline this study measured, crawlability, canonicalization, sitemap availability, and structured data, is not a niche search-optimization concern. It is, according to Google's own stated position, the same baseline that determines eligibility for AI-driven discovery.

Research Question

Are the public-facing websites of leading Indonesian universities built with the crawlability, canonicalization, sitemap, and structured-data signals that Google's own documentation ties to standard Search and AI Overview/AI Mode eligibility?

Methodology

Collection date: 18 September 2026.

Population and sample: the sample was drawn purposively rather than randomly: ten institutions consistently recognized among Indonesia's largest and most research-active universities, spanning public research universities (Universitas Indonesia, Institut Teknologi Bandung, Universitas Gadjah Mada, IPB University, Universitas Airlangga, Universitas Diponegoro, Institut Teknologi Sepuluh Nopember, Universitas Padjadjaran) and large private universities (BINUS University, Telkom University). This is a purposive, not a random or stratified, sample, and the findings describe this sample only; they should not be read as a national percentage.

Collection method: each homepage and each site's publicly accessible robots.txt file was retrieved directly and reviewed for: HTTPS delivery, presence of <link rel="canonical">, presence of <link rel="alternate" hreflang>, presence of schema.org structured data (JSON-LD or microdata), general rendering approach (server-rendered versus JavaScript-dependent), and, from robots.txt, disallowed paths and any declared Sitemap: directive.

Variables: HTTPS status; canonical tag presence; hreflang tag presence; structured-data presence and type where determinable; robots.txt disallow scope; sitemap declaration and validity.

Analysis approach: each site was assessed on each variable from the content actually reviewed; where a variable could not be confirmed from the visible page content, the finding is reported as "not confirmed" rather than as a negative, and the two are kept distinct throughout this article.

Limitations: two of ten sites, Universitas Indonesia at both endpoints and Universitas Padjadjaran at its robots.txt, could not be retrieved for review, most likely due to bot-management or firewall configurations, and could not be assessed technically; this says nothing about their actual implementation. This review examined rendered page content rather than complete raw HTML source in every case, so a small number of "not confirmed" results may reflect incomplete visibility into the page rather than a genuine absence of the feature. The sample is purposive and limited to ten institutions and does not represent the full population of Indonesian higher-education websites. No Search Console impression, click, or position data, and no crawl-log data, were available for this study, so it measures the technical signals a site presents rather than actual indexing or AI-citation behavior. Findings reflect a single collection date and may not reflect current implementation if sites have since changed.

Results

Analysis

The most consistent finding is basic technical hygiene: canonical tags and sitemap declarations are the norm across the reviewable sites in this sample (7 of 8 and 6 of 8 respectively), suggesting that these institutions' web teams, or the content-management platforms behind their sites, several of which are recognizably WordPress-based, handle the fundamentals of duplicate-content signaling and crawl guidance by default. Structured data is the clear gap: only two sites showed any JSON-LD at all, and in both cases the specific schema.org type could not be confirmed, meaning it is not possible from this evidence to say whether either site implements EducationalOrganization, CollegeOrUniversity, Course, or another type Google's documentation associates with rich results.

The hreflang finding is the most actionable one. Every bilingual site in the sample offers an English version, and most signal that fact through an Open Graph og:locale/og:locale:alternate pair or a simple navigation link to an /en/ path. Neither mechanism is the same as the <link rel="alternate" hreflang="x"> tag that search engines use to decide which language version of a page to serve a given searcher. Open Graph tags are built for social-sharing previews, not language targeting. This is a case where an institution may reasonably assume its bilingual content is properly signaled to search engines because a language switcher exists in the visible design, when the underlying markup search engines rely on for that purpose is absent, a gap between what a human visitor experiences and what a crawler can act on.

The single case of a sitemap directive pointing to a placeholder example.com domain, at Universitas Airlangga, is a concrete, low-cost misconfiguration: the file otherwise follows a normal robots.txt structure, and the sitemap line was very likely never updated from a template or default configuration. It has no bearing on the university's actual web presence quality; it is the kind of one-line technical oversight that is common across many kinds of organizations' websites, not particular to this institution.

Two sites being unreachable for direct review, Universitas Indonesia entirely and Universitas Padjadjaran's robots.txt, does not establish that Googlebot or AI crawlers are blocked; different bots are frequently treated differently by firewall and bot-management rules. It does mean these two institutions' actual technical posture could not be verified in this study, and is noted as an open question rather than a finding.

University Examples

Institut Teknologi Sepuluh Nopember's robots.txt stood out for breadth rather than restriction: it declared roughly 105 separate sitemap entries spanning faculties, administrative units, and research centers, suggesting a highly distributed content architecture with sitemap coverage extended to each unit rather than a single institution-wide file. Universitas Gadjah Mada was the only site in the sample where JSON-LD structured data, a canonical tag, and a sitemap declaration were all observed together, making it the most complete technical profile in this sample on the variables measured. This describes only the variables tested here, not a general judgment of the site's quality or of the university.

Implications for Higher Education

For university web and IT teams, the practical gap to close first is hreflang: supplementing Open Graph locale tags with a proper <link rel="alternate" hreflang="id"> / hreflang="en" pair on bilingual pages is a well-documented, low-effort change. For communications and digital teams, structured data represents the largest opportunity in this sample; Google's documentation ties eligibility for enhanced result formats directly to complete, accurate schema markup, and only two of eight reviewable sites in this study had any. For institutional leadership, the finding that two sites were entirely or partly unreachable for external review is worth an internal check with IT, to confirm that legitimate search and AI crawlers, not only human browsers, can reach every public page. For libraries and research offices, this study did not examine research-output pages, ORCID or DOI presence, or repository structured data specifically, a natural next study, and one where schema types such as ScholarlyArticle or Dataset would be the relevant markup rather than the organizational homepage schema examined here.

Recommended Actions

• Audit homepage and key landing pages for <link rel="alternate" hreflang="..."> implementation on every language variant, rather than relying on Open Graph locale tags or navigation links alone.
• Validate any existing schema.org markup with Google's Rich Results Test and confirm the schema type matches the page's actual content, such as CollegeOrUniversity or EducationalOrganization for the homepage.
• Where no structured data exists, evaluate adding EducationalOrganization markup to the homepage and Course markup to program pages, following schema.org's published requirements.
• Check that any Sitemap: directive in robots.txt resolves to a live, self-hosted XML sitemap rather than a placeholder or default value left over from a template or migration.
• Confirm, using the institution's own server or CDN logs and bot-management console, that major search and AI crawlers are not inadvertently blocked by firewall rules; this cannot be verified from outside the network and should be checked internally.

None of these actions is presented as a guarantee of improved search visibility, ranking, or AI citation. Google's documentation ties eligibility to these baseline signals but does not promise any specific outcome from implementing them.

Limitations

This study used a purposive sample of ten institutions rather than a random or stratified sample of Indonesia's full higher-education sector, so its findings describe this sample only. Two sites could not be technically assessed in whole or in part, which itself may indicate bot-management configurations worth an internal review but does not indicate their actual search or AI-readiness posture. This review relied on rendered page content rather than complete raw HTML source in every case, so some "not confirmed" results may reflect incomplete visibility into the page rather than a genuine absence of the feature; this is noted per finding above. No Search Console impression, click, or position data, and no crawl-log data, were available for this study, so it measures technical signals present on the page rather than actual indexing or AI-citation behavior. Findings reflect a single collection date and may change as institutions update their websites.

Conclusion

Among ten leading Indonesian universities sampled in September 2026, canonical tags and sitemap declarations were common, but structured data was rare, present on only two of eight reviewable sites, and a technically correct hreflang implementation was absent from all eight. Two sites could not be assessed at all due to access restrictions encountered during review. Google's own documentation states that AI Overview and AI Mode eligibility rests on the same crawlability and indexing baseline as ordinary Search, which makes these findings directly relevant beyond traditional search-ranking concerns: the gaps identified here are inexpensive to close and do not require any AI-specific technical work, only closer alignment with Google's existing, publicly documented Search guidance.

Sources

• Google Search Central, "Intro to How Structured Data Works" — developers.google.com/search/docs/appearance/structured-data/intro-structured-data
• Google Search Central, "AI Features and Your Website" — developers.google.com/search/docs/appearance/ai-features
• Direct review of publicly available homepages and robots.txt files of the ten sampled university domains, collected 18 September 2026.
• Webometrics.org analytics, 20 June 2026 to 17 September 2026 (Indonesia country-level engagement figures).

Suggested Citation

Webometrics by UNIRANKS (2026). How Discoverable Are Indonesian Universities in AI and Search Results? A Webometrics Structured-Data and Indexability Study. Webometrics.org, 18 September 2026.