On July 14, Phoenix Education Partners reported Q3 2026 earnings, missing revenue estimates by nearly $5 million. On the call, the company told Wall Street the reason: Google AI search algorithm changes hitting January enrollment. Two months earlier, 5W had published research predicting exactly that outcome. The Online Universities AI Visibility Index 2026 called it the Phoenix Paradox.
AI Citation Share — the share of AI-generated answers in which a brand is named or cited — is emerging as a leading indicator of enrollment economics, buyer discovery, and unit economics for any category that has run through Google for the past two decades. Phoenix is one of the first public-company earnings prints to confirm it in the language of a covered quarter.
What the research measured
The Online Universities AI Visibility Index 2026, released in May, ranked the top 25 U.S. online universities by AI citation share across Claude and Google AI Overviews. Thirty-five prospective-student prompts. Two engines. Q2 2026 measurement. Index 1.1, adding ChatGPT, Perplexity, and Gemini, is scheduled for Q3 2026.
The University of Phoenix ranked among the lowest in the category — an estimated 1.5% AI citation share, despite category-leading paid-search spend. The paid-search-to-AI-citation gap was roughly 8-to-1 against Phoenix. The largest such gap the AI Visibility Index series has measured in any U.S. consumer category to date.
The cause is straightforward. AI engines still surface 2010s-era regulatory and accreditation coverage when prompted about online universities. Phoenix reaffirmed its accreditation with the Higher Learning Commission in 2023. Answer engines do not weight publication recency the way human readers do. They weight publisher authority — and the original coverage of the historical controversy lives in high-authority outlets that ChatGPT search, Claude, Perplexity, and Google AI Overviews cite from.
A prospective student in 2026 asking an AI engine for the best online university gets an answer set that does not include the school with the biggest paid-media budget in the category. In many queries, the AI Overview renders above the paid ad slot on Google. The ad is bypassed before it is served.
The earnings call
On the July 14 earnings call, Phoenix Education Partners CFO Blair Westblom guided revenue to the lower part of the range "just given the marketing dynamics that we've observed." The transcript flagged the driver directly: impact from Google AI search algorithm changes affecting January enrollment. The company said it is migrating content to YouTube in response.
Google's algorithm shifts are not the only variable in the Phoenix quarter. Retention timing, B2B channel movement, and broader for-profit enrollment cycles all play roles. But it is meaningful that the company chose AI search as the language for the miss on a public earnings call. It is the first earnings-transcript confirmation of the AI-visibility thesis from inside a covered public company. It will not be the last. Sell-side analysts covering the online education stocks — LOPE, PXED, CVSA, and adjacent names — are already asking questions that map to the Index directly.
The category signal
Online higher education is one of the most paid-search-dependent categories in the American economy. U.S. higher-ed advertising spend reached $2.2 billion in 2019, according to Kantar data reported by The Hechinger Report. Per-enrollment acquisition costs run $2,849 on average, and $5,000 to $8,000 for MBA and executive programs, according to the UPCEA Marketing Survey.
The share of college-bound students using generative AI in their search jumped from 26% in spring 2025 to 46% by late 2025, according to EAB — a near-doubling in eight months. The discovery funnel now starts at ChatGPT search, Claude, Perplexity, or a Google AI Overview. By the time a prospective student runs a paid-search query, the AI engine has already filtered the choice set.
The schools winning that filter are not the schools winning the paid-search auction. Western Governors University, Southern New Hampshire University, and Arizona State University Online together capture an estimated 35% of all AI citations in the Index. Public flagship online divisions — Penn State World Campus, University of Maryland Global Campus, Oregon State Ecampus, University of Florida Online — capture another 23%.
Seven established schools each capture under 0.5% AI citation share. Functionally invisible to the prospective student who begins her search inside an answer engine. When the discovery layer moves and the brand does not appear in the answer, no amount of paid-search spend downstream recovers the impression.
The EdTech parallel — and why it matters for public companies
Higher education is not alone. 5W's EdTech AI Visibility Index 2026 ranked the top 25 online education and EdTech brands by AI citation share across ChatGPT, Claude, Perplexity, and Google AI Overviews. Duolingo, Khan Academy, Coursera, MagicSchool, and Synthesis led. A clear set of legacy SEO winners — Chegg, Course Hero, Quizlet, Study.com — that had dominated Google rankings for a decade all ranked outside the top 15. Several have been the subject of high-profile traffic collapses linked to AI Overviews.
The pattern generalizes. In the U.S. Grocery Retail AI Visibility Index, Walmart — holding roughly 21% of U.S. grocery market share — ranks only fourth in AI citation share, behind Costco, Trader Joe's, and Whole Foods. The largest grocer in the country is not the most-cited grocer in the answer box. In theme parks, Disney captures 58% of AI citations, Universal 23%, Six Flags 13%. Market share and citation share are two different scoreboards, and they are diverging.
Why PR now matters more than SEO
The most consequential shift for communications leaders is not tactical. It is structural.
Earned media has always shaped brand perception. Now it shapes what AI engines cite. Reporting from The New York Times, Bloomberg, Forbes, and trade press becomes training data and retrieval material for ChatGPT search, Claude, Gemini, Perplexity, and Google AI Overviews. Authoritative journalism is compounding inside AI answers with a half-life measured in years, not news cycles. A well-placed Bloomberg article in 2026 will still be surfacing in AI answers in 2030.
That inverts the SEO era's economics. Search engine optimization was a technical discipline aimed at ranking pages against an algorithm. AI visibility is an authority discipline aimed at earning citations from the publisher graph that AI engines already trust. Reputation, once a communications output, is now a retrieval input.
The Phoenix Paradox is a case study in this shift. Fifteen years of paid-search performance did not translate into AI citations. Fifteen-year-old regulatory coverage did — because it was published in authoritative outlets that AI engines weight heavily. Communications built the visibility. Communications will have to rebuild it.
What comes after the search box
The playbook for defending citation share inside AI engines has four components. Each looks familiar. Each has a different weight than it did five years ago.
First, earned media programs that target the publisher graph AI engines cite. Not media impressions. Citations inside the outlets — trade press, federal data sources, category rankings, and community platforms — that answer engines pull from. For online higher education, that means U.S. News, Forbes Advisor, Niche, BestColleges, NCES/IPEDS, Nurse.org, and Reddit. For every other category, the publisher graph is different but the principle is the same.
Second, Generative Engine Optimization — the technical and editorial discipline of earning citation inside answer engines. Schema, entity clarity, structured content on the exact prompts prospective buyers ask, first-party publication of the facts AI engines need to summarize a brand accurately.
Third, an AI Citation Audit — a diagnostic that measures the brand's current standing across ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews before any remediation begins. Where is the brand cited today? Which competitors capture more citation share? Which sources feed the answer? Where is the recoverable ground?
Fourth, a monthly cadence. Categories move quickly. AI citation share is not an annual metric. Online higher education is among the fastest-moving consumer-citation categories 5W has measured. A quarterly audit would have missed the Q1 2026 inflection that surfaced in the Phoenix earnings call three quarters later.
The broader lesson
Every category that runs on paid discovery is on the same clock as online higher education. Insurance. Financial services. DTC retail. Cosmetic surgery. Home services. Legal. The question is the same in each: when the discovery layer moves from the search box to the answer box, does the brand appear in the answer?
The companies that measured search rankings for two decades now need to measure AI visibility. Markets eventually price what customers discover.
Phoenix just showed the market what the cost of inaction looks like. Every category is next.
Ronn Torossian is the founder and chairman of 5W AI Communications, the AI Communications Firm. He is the publisher of Everything-PR and the author of two best-selling editions of For Immediate Release. Related reading: Why Citation Share Is the New Market Share and Reputation Is Now the Paragraph the Machine Writes.