Field notes from the new front door of higher ed.
AI search has rewritten how prospective students find a school. Our team publishes what we’re learning from live audits, citation studies, and 90-day GEO engagements with colleges and universities across North America.
A strong Canadian university is nearly invisible in AI search.
We ran a live SEO/AEO/GEO audit on a top-ranked Canadian school. The site reads beautifully to a human reader, and almost not at all to a language model. Here is what we found — and the five gaps we see at almost every Canadian institution.
AI search is higher ed's new front door.
Prospective students no longer start on a school's website. They start with ChatGPT, Perplexity, Gemini, or an AI Overview. If the engines aren't citing you, the campus tour never gets booked.
Reddit is winning the college comparison query.
When a student asks an AI engine to compare two schools, the model needs a source. Most universities have a policy of not naming rivals. Reddit does not have that policy. So a 2021 thread becomes the canonical answer.
Rankings reward inputs. AI rewards being the answer.
A school's national ranking and its citation share in AI answers are now two different rankings — and they are pulling apart. A mid-sized school can out-cite a U15. The frame inside enrollment leadership has to shift.
JSON-LD for universities: a field guide.
The mechanical work that moves AI citation share more than anything else is schema. Here is a practical, opinionated guide to which schemas to ship, in what order, on a university website — with code you can paste in this week.
llms.txt: the robots.txt for the AI era.
A small file at the root of your domain that tells AI engines what your site is, where the canonical content lives, and what to read first. Cheap to ship. Disproportionate effect on citation share.
Wikipedia and Wikidata: the hidden authority layer behind every AI answer.
AI engines lean on Wikipedia. They lean on Wikidata. A stale article about your institution is silently weakening every AI answer about you — and the fix is harder than people expect, but high-leverage when done right.
The 2027 enrollment cliff is real. Marketing spend should change shape.
U.S. high-school graduating classes will shrink by ~13% between 2025 and 2041. Canada faces a different demographic mix shift. Most schools are budgeting for the old curve. Here is how spend should change — by channel, by region, by audience.
The satellite-campus blindspot in AI search.
Most universities have a beautiful new urban campus that doesn't appear in AI answers about its own city. The problem is the same at every second campus. The fix is mostly mechanical, and is among the highest-converting work we do.
A citation audit of ChatGPT, Perplexity, Gemini, Copilot, and AI Overviews.
We tested 200 higher-ed queries across five major AI surfaces. Here are the citation patterns each one rewards, the kind of source each one prefers, and what that means for where to invest your AEO effort.
FERPA-compliant AI agents: a buyer's guide for higher ed.
Half the AI vendors pitching universities aren't actually FERPA-compliant. Most haven't even read the regulation. Here are the eight questions an enrollment leader, registrar, or CIO should ask before signing.
A 90-day GEO plan for universities you can execute this quarter.
The fastest path from "we don't show up in AI answers" to "we are cited in 8 of 15 test queries." Three sprints, thirteen weeks, measurable lift at the end. Here is the plan we run for every new institutional engagement.
The 2.8 million-student collapse no one is marketing around.
The enrollment cliff isn't a future problem. It's a past-tense one — and almost all of it has already hit the community-college tier. Two-year UG enrollment is down 44% in fourteen years. Most four-year transfer-pipeline math is now broken; almost no one has rebuilt it.
The selectivity myth: 92% of US four-year colleges shouldn't be marketing on it.
The Scorecard says one in five U.S. four-year colleges admits more than 90% of applicants, and the typical four-year admits 77%. Past the 25% admit-rate line, median graduate earnings are essentially flat. Selectivity theater is now an SEO and brand liability for almost every institution.
Sticker shock is a top-of-funnel killer. Here is the actual math.
The single most damaging piece of misinformation in the higher-ed funnel is the published sticker price. Median net price for a low-income family at a private nonprofit is $17,887 — half the sticker. Almost no school surfaces that number before the financial-aid letter, by which point the prospect has already eliminated them.
The hidden-ROI institutions the market mis-prices.
Sort the College Scorecard by earnings-to-debt ratio and the top of the list contains the names you'd expect — and several you wouldn't. Federal service academies. Specialized health-sciences institutions. Pharmacy schools. Each one out-earns most Ivies and almost no prospective family has heard of them. The marketing leverage is enormous.
An HBCU AI-search audit keeps finding the same five issues.
We've now audited a half-dozen Historically Black Colleges and Universities for AI search visibility. The institutions are extraordinary — Research 2 designations, NASA partnerships, century-old prestige. The digital surface is consistently the weakest part of the operation. Here is the pattern, with the institution kept anonymous.
International recruitment after the Canadian study-permit cap.
The 2024 and 2025 Canadian study-permit policy changes have rewired international student flow across North America. Indian and Vietnamese demand is still strong; Chinese pipeline is in long-run decline; Nigerian and West African pipelines are emerging. Here is the diversification playbook for the rest of the decade.
The 2026 higher-ed conference circuit: where the leaders actually meet.
The honest, opinionated guide to which 2026 higher-ed conferences are worth a senior marketer's time, which are worth an enrollment leader's, and which are worth skipping. Built from our own outreach data across roughly 1,500 contacts at 27 conferences.
The master's debt trap: which fields still pay the math.
The undergraduate debt narrative in the U.S. is broadly miscalibrated. The graduate-school narrative is the opposite. Here is the field-by-field breakdown of which master's programs still pay back the borrower, and which are structurally fragile — from the College Scorecard's field-of-study cohort.
The transfer-student funnel: 38% of degrees, 4% of budgets.
Roughly 38% of U.S. bachelor's degrees are earned by students who started at a different institution. The marketing budget allocated to capturing them is closer to 4%. Here is what the under-investment is costing — and what a serious transfer funnel looks like in 2026.
Every retention point is worth $40 of CPC.
Median full-time freshman retention at U.S. four-year institutions is 76.8%. Median 6-year completion is 55.5%. The unit economics of recruitment quietly degrade each step down the funnel — and the dashboards in most enrollment shops only watch the top.
The Pell-by-field equity story: where credentialing closes the gap.
The earnings gap between Pell and non-Pell graduates of the same major varies enormously by field. In nursing it's $662. In computer programming it's $15,586. The College Scorecard's field-of-study cohort tells you exactly where to direct equity-focused enrollment work — and the pattern is operationally usable.
AI agents inside admissions: a working architecture.
What it actually looks like to deploy an AI agent inside an admissions office in 2026 — what the agent does, what stays with humans, where the audit trail lives, what the staffing model becomes, and what the failure modes are. Built from real deployments across our agent catalog.
Find out what AI says about your school.
Paste any program page. Get a structured SEO, AEO, and GEO audit in 30–60 seconds. Free. Instant. Built by Ibex Insights.