A real audit. A real institution. We don't name it.
A nationally-ranked HBCU. Research 2 designation. Multiple federal research partnerships across the science agencies. A double-digit enrollment surge in the post-pandemic period. A specialized natural-sciences research capability few peers can match. A historically significant on-campus collection.
The institution is genuinely extraordinary. The digital surface is the weakest part of the operation by a wide margin. We've now run roughly the same audit on five other HBCUs. The pattern is consistent.
The five issues we keep finding
1. Site speed catastrophic enough to fail crawlers entirely
The flagship homepage on this audit took 30+ seconds to load, and timed out for several major AI crawler user agents in our test runs. A site the bot can't reach is a site that doesn't exist in the answer. Every other AEO improvement on the list is downstream of this one.
This is the single most-common HBCU-specific issue we see. The infrastructure has been deferred. The CMS is often a 2010-era WordPress with twelve plug-ins. The CDN is unconfigured or absent. The fix is real engineering work — not a content sprint — and it has to come first.
2. No detectable schema markup
Across the program catalog: zero EducationalOccupationalProgram blocks. No Course. No FAQPage. No Person for named faculty. The AI engines that read structured data to build answers have nothing to read.
The fix is mechanical. We've shipped templated EducationalOccupationalProgram across full HBCU program catalogs in two engineering weeks. The before-and-after on AI citations is measurable in 30–45 days.
3. Content not structured for AI consumption
Long marketing pages with a single H1 and twelve paragraphs of brand copy. No Q&A. No transparent comparison content. No machine-friendly tables of cost, duration, or outcomes.
The page reads beautifully to a person. To an AI engine assembling an answer about "best HBCU in [state] for [program]", the page is undifferentiated text. The model goes elsewhere.
4. Outcomes documentation that loses every comparison
This one is HBCU-specific and important. The most common AI-driven question about an HBCU is some variant of "which HBCU is best for X." The model needs to compare. Every HBCU's Scorecard data is public — completion rates, median earnings, median debt, Pell share. Most HBCU websites do not contextualize their own numbers; they publish the headline and stop.
The result is a comparative narrative the institution does not control. A 56% completion rate published without context loses to a 77% completion rate published without context, every time. That is not the right comparison — the institutions serve very different populations — but the AI engine doesn't know that unless the institution publishes the context.
5. Faculty research visibility gaps
This HBCU has world-class researchers across several natural-sciences and engineering fields, with active federal-agency research partnerships. Almost none of those faculty have a complete Person schema block on their page. Most do not have ORCID linked. Most do not have Google Scholar linked. Wikidata items, where they exist, are stale.
The result: when an AI engine assembles an answer about "top researchers in [field] at HBCUs," the institution's actual leaders in the field are invisible to the engine, and competitors with better-tagged faculty pages get cited instead.
The HBCU brand is, on average, stronger than the HBCU digital surface. The gap is the entire opportunity.
What an HBCU-specific 90-day plan looks like
The plan is the same as the general 90-day plan we publish, with three HBCU-specific overlays:
Overlay 1 — Site speed first
Most HBCUs cannot proceed with the standard schema-and-content plan until the site itself loads. We typically front-load engineering work for the first three weeks: CDN setup, image optimization, plug-in pruning, server-rendered HTML for tables that were JavaScript-rendered. The schema work starts in week four.
Overlay 2 — Outcomes context, not just outcomes
HBCU outcomes data has to be published with the population context. The page that says "56% 6-year completion, 67% Pell-eligible, 39% first-gen, 30% lower COA than the national median" tells a coherent story about an institution serving the highest-need cohort in the country at a structurally lower price. The page that just says "56% 6-year completion" loses the comparison.
This is one of the few places where AEO and equity strategy align perfectly. The honest contextual presentation of the numbers is also the presentation that wins AI citations.
Overlay 3 — Category claims that are credibly defensible
The HBCU we audited can credibly claim several things along these lines:
- A regional best-in-class category claim — "the strongest HBCU in [region] for [program]" with sourced evidence.
- A single-rare-program claim — a sport, a degree, a research center the school is unique on.
- A federal-partnership claim — "one of the few HBCUs with [agency] research partnership."
- A historically-significant-asset claim — a collection, a building, a piece of campus history with national importance.
Each of those is a category claim with structured evidence behind it. None of them was prominent on the homepage at the time of the audit. Each becomes a category page with sourced proof, internal links, and a clear positioning statement that AI engines can pattern-match to.
The overall principle: lean into the specific, defensible, sourced claim. Generic mission-statement copy does not surface in AI answers. The category page that says "the only HBCU in [region] with [unique research partnership]" does.
Why this matters more for HBCUs than for the average institution
Two structural reasons:
The Reddit problem is sharper at HBCUs
The most-cited Reddit threads about HBCU choice are old. They reflect a specific moment in HBCU enrollment that has changed materially since 2020 (the pandemic-era HBCU enrollment surge, the post-2023 racial-conscious-admissions decisions in U.S. higher ed, the new Research 2 and Research 1 designations several HBCUs earned in 2025). The model uses what's available. What's available is often years stale and out of context.
The institutional brand has more to lose
An HBCU's brand promise is built on a century-plus of mission and outcome. When an AI answer mis-summarizes the institution — citing 2018 enrollment numbers, missing the Research 2 designation, missing the NASA partnership — the brand cost compounds.
What to do this quarter if you lead an HBCU
- Run a citation audit across ChatGPT, Perplexity, AI Overviews, and Gemini for fifteen branded and comparison queries. Snapshot what the engines say about you today.
- Run a Lighthouse and crawler-readability test against your homepage and your top five program pages. If load times exceed three seconds, that's the work-zero item.
- Pull your Scorecard data and identify three category claims you can defensibly own. Build a category page per claim.
- Audit your top thirty named faculty pages for ORCID, Google Scholar, and
Personschema completeness. - Identify three to five comparison queries — your school vs. peer HBCUs — where Reddit is currently the canonical source. Ship comparison pages that rebuild the answer.
The HBCU sector has done extraordinary work over the last five years. The digital surface has not yet caught up to the institutional reality. Closing that gap is the single highest-leverage marketing opportunity in the segment.
If you lead an HBCU and want a private audit run on your institution, the first one is free. We anonymize results.