When a prospective student wants to know which European MBA has the strongest reputation, or which school is best for a career in consulting, a growing number of them no longer open Google and work through a page of links. They ask ChatGPT, Gemini or Claude, and they read the answer they’re given.
That answer is rarely a neutral list. It is a recommendation, a comparison, a ranking, or a short summary of what a school is known for. And it is assembled at speed, from sources across the open web, often before a candidate has visited your website or spoken to your admissions team.
This is the shift we wanted to understand properly rather than guess at. So, earlier this year we began a structured study of how the major language models talk about business schools, and we’ll publish the full findings in November. This post explains what we are doing, why it matters for comms teams specifically, and how you can see the results first.
Most of the conversation about AI and search has been generic. The reason it deserves a closer look in our sector is that business education is a comparative market, and comparison is precisely what these models are built to do.
Candidates rarely consider a school in isolation. They weigh schools against one another by geography, ranking position, employer links, return on investment and reputation. Those are exactly the kinds of questions large language models (LLMs) answer well, because they are comparative and advice-led.
The wider data already tells us where this is heading. ChatGPT reported around 900 million weekly active users at the start of this year, up from 400 million a year earlier (OpenAI / TechCrunch, 2026). In 2024, Gartner forecast a 25% drop in traditional search engine volume by 2026, a prediction that looked bold when it was made and now looks conservative, and SparkToro found that 68% of Google queries already end without a click on any result in what is coined as zero-click search. Survey work published by the Harvard Business Review found that two in three young adults had used an AI tool to replace a Google search in the previous month (Harvard Business Review, 2026). The effect is already visible in our own sector: research found that among prospective students who use AI in their search, around a third had added a school to their shortlist because of an AI answer, and nearly one in five had been steered away from one (EAB, 2026).
For a business school, that means your reputation is now being interpreted by tools you don’t control, drawing on sources you didn’t necessarily choose, and presented to a prospective student as a single confident answer.
We have built this as a repeatable visibility tracker rather than a one-off snapshot, because a single moment in time would tell you very little about something moving this quickly.
We are testing five major models, ChatGPT, Gemini, Claude, Grok and DeepSeek, against a set of business-school prompts designed to mirror the questions real candidates ask. We’re working towards a target of over 1,000 full responses, collected over a six-month window. Because each response can mention several schools, cite multiple sources and structure its rankings differently, the dataset sitting beneath that number is considerably larger.
For every response, we record which schools are mentioned, the order they appear in, whether they are ranked, recommended or only referenced, and every source cited down to the individual webpage. We are deliberately careful with those distinctions, because a mention is not a recommendation, and a recommendation is not a ranking. Treating them as the same thing would overstate what the models have actually said.
To keep the research honest, the full raw output is preserved, every entry is reviewed by a person rather than a script, and we cap how many times the same model and prompt combination can be logged in a day, so no single moment skews the picture.
We are still collecting data, with the final set due in September, so we are holding most of the detail back for the November report. But one pattern is already clear enough to share.
The sources these models lean on most heavily are not the ones schools own. Earned media is doing a great deal of the work, which echoes what broader industry research is also finding across sectors: Muck Rack’s analysis of more than a million AI citations found that 84% came from earned media sources (Muck Rack, 2026). The encouraging part, and the part most relevant to anyone reading this, is that the majority of the sources we are seeing cited are ones a comms team can influence. Your media coverage, your rankings commentary, your owned content and your presence on third-party platforms are the raw material these tools are reading.
What it means in practice, and the specific figures behind it, is what the November report will set out in full.
The full report lands in November, with a much richer picture of which schools are appearing, what they are being associated with, and which sources sit behind those answers. It will give comms teams a way to pressure-test reputation: to move from asking whether you appear in AI answers to understanding what you appear for, where that comes from, and whether it reflects the reputation you are working to build.
If you would like early access ahead of publication, email Stephanie Mullins-Wiles and we will make sure the findings reach you first.
Author: Jennifer Wright
BlueSky Education's Marketing Director is a seasoned marketing professional, experienced in content marketing, social media, engagement strategy, CRM management, marketing measurement (ROI), SEO, AEO and GEO.