GEO Monitoring: Track Your School's Visibility in AI Answers

Why GEO monitoring is now essential for universities

Optimising your presence in AI engines without measuring it is like running a recruitment campaign without tracking applications. GEO — Generative Engine Optimisation — produces measurable results, but only if you have a tracking system to identify what works and what remains invisible.

In the UK, ChatGPT mentions a university in just 29% of higher education responses. Perplexity reaches 38%. The European average sits at 19% (Source: Skolbot GEO Monitoring, 500 queries x 6 countries x 3 AI engines, Feb 2026). Your institution is absent from more than seven out of ten AI responses. Without monitoring, you will never know whether your GEO efforts are changing that reality.

For the foundations of GEO and why it matters for your institution, see our comprehensive GEO guide for schools.

What GEO monitoring actually measures

GEO monitoring is not simply checking whether your university "shows up in ChatGPT." It is a structured tracking system built around three distinct metric families.

Citation rate

Citation rate measures how often your institution is named in AI responses for a predefined set of queries. You submit the same 20 to 50 queries each month — "best business school in London," "MBA finance UK AACSB accredited," "engineering degree apprenticeship West Midlands" — and count how many responses mention your institution.

This rate must be calculated per AI engine (ChatGPT, Perplexity, Gemini) because results vary substantially between them. A citation rate of 42% on Perplexity and 18% on ChatGPT for identical queries is a common pattern that directly shapes your strategy.

Attribution rate

Attribution goes beyond mere mention. It checks whether the AI engine cites your source — a link to your website, a reference to a specific page — or simply names your institution without directing traffic to you. On Perplexity, sources are displayed systematically. On ChatGPT, attribution is less frequent but measurable through in-response links.

Mention context

Context qualifies the nature of the citation. Is your university mentioned as a top recommendation, as an alternative, in a list, or with caveats? "Imperial College London is the leading choice" and "other institutions such as X also offer..." carry very different recruitment value.

Tools for setting up your GEO monitoring

Method 1: structured manual auditing

The most accessible method requires no paid tooling. Create a spreadsheet with your target queries (30 to 50 queries covering your programmes, location, and specialisations), then submit them manually to ChatGPT, Perplexity and Gemini once a month.

For each response, record: mention (yes/no), position in response (1st mention, 2nd, 3rd+), attribution (link yes/no), and context (recommendation, alternative, list). This work takes approximately 3 hours per month for 40 queries across 3 engines.

Method 2: API-driven monitoring

Perplexity offers an API that automates query submission and returns structured responses including cited sources. The API delivers citations with their URLs, enabling automatic calculation of your attribution rate.

For ChatGPT, the OpenAI API with the GPT-4o model and web_search parameter enabled replicates the search engine behaviour. Cost remains modest: approximately £12 per month for 200 weekly queries.

Method 3: Skolbot AI Check

The Skolbot AI Check tool runs an AI visibility audit in minutes. Enter your institution name and target queries, and the tool interrogates the major AI engines to produce a structured report: citation rates, cited sources, and improvement recommendations. It is the fastest entry point for a first diagnostic.

Building your GEO dashboard

An effective dashboard tracks evolution over time rather than capturing a single snapshot. Here is the structure we recommend for institutions:

Update this table monthly. The "Change vs prev. month" column matters most — it reveals the actual impact of your GEO actions on real visibility.

The recommended monitoring cadence

Weekly: spot-checks

Each week, submit 5 to 10 strategic queries to ChatGPT and Perplexity. The goal is not statistical rigour — it is early detection. If your institution suddenly vanishes from a query category, you catch it within 7 days rather than 30.

Monthly: full audit

Once a month, run your complete query battery (30 to 50) across all three AI engines. Update your dashboard, calculate month-on-month changes, and identify which query categories are improving or declining.

Quarterly: strategic review

Each quarter, analyse trends. Compare your citation rate against direct competitors (by submitting the same queries). Adjust your query list to reflect new programmes, updated THE or QS rankings, UCAS changes, and market shifts.

How to interpret results and take action

Scenario 1: low citation rate across all engines

Your institution lacks foundational signals. The priority is implementing Schema.org structured data — institutions with structured Schema.org markup gain an average of +12 points in GEO visibility (Source: Skolbot GEO Monitoring, Feb 2026). This is the fastest lever available.

Scenario 2: strong on Perplexity, weak on ChatGPT

Perplexity relies more heavily on real-time web content; ChatGPT draws primarily from its training corpus. This gap means your web content is solid but your institution lacks presence in the sources ChatGPT weights most: rankings, UCAS profiles, QAA reviews, THE features. Invest in your presence on these third-party platforms.

For a deeper analysis of what ChatGPT cites and why, read our guide on content cited by ChatGPT for schools.

Scenario 3: listed but never first

Your institution is known to the AI engine but not perceived as a leader. Strengthen authority signals: verifiable accreditations (AACSB, EQUIS, TEF Gold), sourced employment outcomes, and precise quantitative data on your programme pages. Russell Group institutions have a natural advantage here, but specialist providers can outperform on niche queries.

Scenario 4: citation without attribution

The engine names your institution without linking to your site. Verify that your pages are crawlable by AI engines (no robots.txt blocks for GPTBot or PerplexityBot), that your Schema.org includes the canonical URL, and that your content is directly accessible in HTML — not buried in PDFs or iframes.

Monitoring competitors to benchmark your progress

Monitoring your own institution alone is insufficient. The same queries submitted to the same engines also produce data on competitors. Systematically record which institutions are cited instead of yours, in what context, and at what frequency.

This competitive intelligence reveals which institutions are investing in GEO. If a competitor jumps from 0% to 28% citation in two months, they have restructured their content. Analyse their site to identify what they changed, and adapt your own strategy.

To specifically audit your Perplexity presence, see our Perplexity visibility audit for schools.

Common GEO monitoring mistakes

Testing once and drawing conclusions. AI responses vary between sessions, depending on conversation context and model updates. A single check has no statistical value. Only trends across multiple months matter.

Choosing overly generic queries. "Best university in the UK" is a query where only 3 or 4 institutions will ever be cited. Target queries specific to your programmes, location, and specialisations — that is where GEO offers the most leverage.

Ignoring mention context. Being cited in "...but other, less established institutions also offer..." is not a win. Mention context is as important as the mention itself.

Confusing AI visibility with web traffic. GEO monitoring measures your presence in responses, not clicks to your site. The two are related but distinct. The attribution rate (presence of a link to your site) is the metric that bridges AI visibility and student acquisition.

FAQ

How many queries should we track for reliable GEO monitoring?

A minimum of 30 queries covering your main programmes, geographic catchment, and accreditations. Ideally 50, split across categories (programmes, geography, specialisations, accreditations, competitor comparisons). Below 20 queries, statistical variation makes trends unreliable.

Do AI engine results change frequently?

Yes, substantially. ChatGPT updates its corpus several times a year and responses vary by conversation context. Perplexity uses real-time web data and can shift results within days. This is precisely why regular monitoring matters — a snapshot tells you almost nothing.

Is paid tooling necessary for GEO monitoring?

No. Manual monitoring remains effective for institutions starting out. A spreadsheet and 3 hours per month are sufficient for structured tracking of 40 queries. Paid tools (Perplexity API, specialist platforms) become relevant when tracking more than 100 queries or systematically benchmarking multiple competitors.

Does GEO monitoring replace Google Search Console tracking?

No. Google Search Console measures your visibility in traditional search (blue links). GEO monitoring measures your presence in generative responses. The two are complementary: an institution can rank well in SEO and be entirely absent from AI answers, or vice versa. Track both.

How do we connect GEO monitoring to enrolment outcomes?

Cross-reference your GEO citation rate evolution with acquisition data. If your citation rate rises from 15% to 30% and organic applications increase over the same period, the correlation is strong. For more precise tracking, add UTM parameters to pages AI engines cite and measure referral traffic from perplexity.ai and chatgpt.com in your analytics platform.