2 of 11
Eleven domains were cited in Madrid across four runs. Two university sites appeared, each in one run only, and neither in the stable set.
Higher education
AI is already answering your applicants and it is not reading your site.
Before they request a prospectus, students and their families ask an assistant where to study. EchoWi runs those questions repeatedly, with country and language fixed, and shows you which institutions get named, which sources the answer is built from, and which recruitment markets you disappear in.
AI answer
Sources it was built from
- Student accommodation sites 4 of 4
- A course guidance platform 4 of 4
- Instagram 3 of 4
- The best-placed university site 1 of 4
What we measured
We asked for the best universities in Madrid and in Paris
Google AI Overview, country and language set explicitly on every run. Five runs in Madrid, three in Paris.
0 of 3
Stable sources that were universities
The three domains present in all four runs are student accommodation and a course-guidance platform. No university among them.
4 of 4
Runs that named the university tracked
One private university in Madrid was named in all four runs while its own site was never cited. Named every time, cited never: two different things.
3 of 4
Runs that cited a social platform
Instagram was cited in three of the four runs. No university site was cited in more than one.
Google AI Overview, Madrid market, Spain and Spanish, re-measured 7 August 2026 over four runs. Some institutions in this data are EchoWi customers; the results are published exactly as they came back. Read the Madrid and Paris measurement.
Why you get named and not cited
Universities publish syllabuses and admissions pages. Neither compares, and the question that decides an enrolment compares.
- The answer comes from the businesses that live around the degree: halls of residence, tutoring firms, guidance platforms and directories. They publish rankings, comparisons and guides, which is exactly what the answer needs to lift.
- Paris repeats the pattern with different players: a language school, a relocation agency and a tutoring firm among the stable sources.
- When a university site does appear in Paris, it appears once in three and split across different subdomains, which fragments the signal instead of accumulating it.
- Being named without being cited is fragile. It rests on what third parties say about you rather than on anything you publish.
Named in the answer, absent from its sources
The most actionable thing a measurement can tell you is that AI cited a source your name appears in and still left you out of the answer. That happened in Madrid: one of the three cited sources opens its article by naming a specific university, and that university is not in Google's answer. The name was in front of the model and did not survive the last step. That rules out crawling, rendering and relevance in one move, and leaves a content problem, which is the only one of the four you fix by writing.
Four ways a family asks the same question, four different source lists
A prospective student does not ask your question once in your words. We asked the same enrolment question four ways in Madrid and four ways in Paris, twice each, and counted the domains that came back every time against those that appeared once and never again.
- Sources cited
- Came back in every run
Madrid, which are the best 7/18
Madrid, where do I study 9/21
Madrid, best reputation 9/16
Madrid, best private universities 9/19
Paris, which are the best 10/19
Paris, where to study 10/33
Paris, best reputation 7/16
Paris, best universities 8/25
| Question, as asked | Sources cited | In every run | Once only |
|---|---|---|---|
| Madrid: which are the best private universities in Madrid | 18 | 7 | 11 |
| Madrid: where do I study in Madrid at a private university | 21 | 9 | 12 |
| Madrid: which private university in Madrid has the best reputation | 16 | 9 | 7 |
| Madrid: best private universities in Madrid | 19 | 9 | 10 |
| Paris: which are the best universities in Paris | 19 | 10 | 9 |
| Paris: where to study in Paris | 33 | 10 | 23 |
| Paris: which Paris university has the best reputation | 16 | 7 | 9 |
| Paris: best Paris universities | 25 | 8 | 17 |
Google AI Overview, AI Mode and Gemini, measured 10 August 2026, two runs per wording, Spain in Spanish and France in French. Prompts are shown translated; they were asked in the market language. The "once only" column is the one to read: across the eight, more than half of everything cited appeared in a single run and never again. A rank you hold in one wording is not a rank you hold.
What an admissions and marketing team gets
A baseline you can take to a rectorate or a board, not a score nobody can explain.
The five questions behind an enrolment
Discovery, the comparison against the institution down the road, the cost objection, employability, and the same question in the language of your recruitment market. None of them contains your name.
Every recruitment market run separately
Country and language fixed on each run. Measuring your Latin American recruitment from Madrid gives you a reassuring picture of somewhere else.
The sources, not just whether you appear
Which domains build the answer, so you know who to be published alongside. In this sector it is almost never another university, and it is often a hall of residence or a guidance platform.
Repeated, so the number means something
Answers vary between identical requests. One check cannot tell a stable citation from a coincidence, which is why every figure on this page has a run count next to it.
We check the answer layer exists in your recruitment markets first
AI coverage is not universal. Measuring five categories in Spain in August 2026, two returned no AI Overview at all across thirteen runs, with no errors. Universities do get one, in Spain and in France, which is why this page exists. We still run the check before quoting anyone, and if a recruitment market of yours has no answer layer we say so rather than selling you an empty dashboard.
How we work in this sector
We work with private universities in Spain and France. The figures above come from those markets and are published unedited, including the part where no university website reached the stable set.
What admissions teams ask us first
The questions that come up in the first call, answered here so you do not have to book one to get them.
Why does AI name our university but never cite our website?
Because the question that decides an enrolment is comparative, and your site does not compare. Admissions pages, syllabuses and news items describe one institution; the answer needs a source that ranks or contrasts several. In our Madrid measurement the three domains cited in every run were student accommodation and a course-guidance platform, and no university site reached the stable set, while one private university was named in all four runs. Being named rests on what other people publish about you; being cited rests on what you publish.
Does this replace our rankings work, like QS or Shanghai?
No, and it measures something the rankings do not. A ranking position is a score awarded once a year by an institution with a published methodology. This measures which sources an assistant actually assembles an answer from, on a given day, in a given country and language. A university can hold a strong ranking position and still be absent from the sources of the answer a family reads first.
How many recruitment markets can you measure at once?
As many as you recruit in, with each run pinned to a country and a language rather than averaged. That separation is the point: a question asked in Spanish from Spain and the same question asked in Spanish from Mexico are different measurements, and a single global number hides exactly the market where you have a problem.
We are a business school rather than a university. Does the same apply?
It applies, and the first thing to check is different. In our August 2026 sweep of Google results pages, "best business school in Europe" returned an AI Overview while "best business school in Barcelona" returned none at all. So for a school recruiting internationally there is an answer layer to compete in, and for the same school asked about at city level on that day there was not. Which of your questions has an answer is the measurement to run before any content work, because optimising for a box Google is not drawing is optimising for nothing.
Can you show that fixing something changed the answer?
Not as a causal claim, and we will not pretend otherwise. What we can do is freeze the question set, measure again on the same surfaces and markets, and show you what moved. Nobody in this category has published a controlled before-and-after with a comparison group, ourselves included, so any vendor promising attributed uplift is describing a study that does not exist yet.
What does the first measurement actually deliver?
Your enrolment questions run repeatedly on the surfaces that answer them in each recruitment market, with three things listed: which institutions were named, which domains built the answer, and which of those domains held across every run rather than appearing once. That last split is what separates a source worth pursuing from noise, and on our own university data more than half of everything cited appeared once and never again.
See what AI says about your institution
Send us your site and the recruitment markets you care about. We run your enrolment questions repeatedly on the surfaces that answer them, and come back with the institutions named, the domains the answer was built from, and which of those held across every run. Country and language are fixed per market, so you see the market and not an average of all of them.
Get your enrolment questions measured (opens in a new tab)No card. If a recruitment market of yours has no answer layer yet, we tell you and we do not bill for watching it.