The classic layer: what still earns its budget
Program-level SEO remains the workhorse: each degree program needs its own page answering cost, outcomes, format, and admission requirements directly, because that is how prospects search and how counselors compare. Paid search on program plus intent terms captures late-stage demand, but it inflates fast in competitive programs, so cap it against enrollment value per program rather than inquiries. Email nurture built around decision milestones, application deadlines, aid dates, deposit windows, outperforms generic newsletters everywhere it is tried. And student-generated proof, outcomes data, real employment numbers, named alumni stories, is the trust layer every other channel borrows from.
What quietly stopped working: generic brand campaigns aimed at rankings-page prestige, and gated viewbooks nobody downloads. The eighteen year old you want does not fill out forms to read marketing; they ask questions and expect answers.
The 2026 layer: students ask AI first
Before a prospect ever hits your .edu, there is now a conversation you cannot see: is this degree worth it, which schools are strong in this field for someone like me, what are my chances with these stats, is campus life good there. ChatGPT, Gemini, Claude, and Perplexity answer with named institutions, synthesized from program pages, outcome data, rankings coverage, forums, and review threads. Parents run the same conversations about cost and safety. If your programs are absent or misdescribed in those answers, you lose applicants who never appear in any funnel report, the exact upstream pattern from what you are being taught about AI visibility is wrong.
The fix list is concrete: program pages that answer the questions assistants get asked, first paragraph, plain language, real numbers. Consistent facts across your site, the common data sets, and directory profiles, because tuition contradictions read as unreliability. Structured data for courses and organization. And presence where students actually discuss schools, which assistants read heavily for the campus-life questions your marketing cannot credibly answer.
A 90 day plan for an enrollment team
Days one to fifteen: baseline. Run your top programs through the four assistants with real prospect questions, best X programs for Y, is X university good for Z, and log who gets named; the free GEO Grader automates the brand-level view. Audit your ten highest-value program pages against the direct-answer standard. Days fifteen to fifty: fix the input layer program by program, answers, schema, fact consistency, outcome numbers published plainly. Days fifty to ninety: extend to the trust layer, respond in the forums where your school is discussed, get outcome stories into the publications assistants cite, and re-run the baseline. Enrollment cycles are long, but assistant answers move on model refresh cycles, so you will see visibility movement inside the quarter even though applications lag it.
Measuring what matters in the new funnel
Higher ed attribution was already murky; AI referrals make last-click reporting fiction. Track three lines instead: share of assistant recommendations for your priority programs, trending monthly, the method in AI search tracking; branded search and direct arrivals to program pages, which is where AI-referred prospects surface, per AI traffic analytics; and stealth applicants, students who apply having never entered your funnel, whose rising share is itself evidence the decision moved upstream. Put those next to cost per enrolled student by channel and the budget conversation changes from impressions to enrollment.
