How AI recommends academic program pages
ORIGINAL RESEARCH • AI SEARCH IN HIGHER ED
What 1,350 AI answers reveal about nursing, business and psychology program pages, and the six changes that help AI quote your school accurately.
Patrick Powers, Powers Digital • Published November 2026 • Data collected October 22-23, 2026
1,350
AI answers from ChatGPT, Perplexity and Google AI mode
98
program pages audited across 35 colleges and universities
0.68
correlation between page completeness and own-site citation
80%
of head-to-head comparisons name no winner
A school's reputation decides whether AI names it. Its program page decides whether AI quotes the school or someone else, and whether what AI says is current.
Key Findings
Being named tracks brand, not page quality. Elite business schools are recommended constantly, but their own pages back up those mentions as little as 4% of the time.
Complete program pages get used. The more of a student’s checklist a page answers, the more often AI cites the school's own site.
AI rarely picks a winner. Comparisons weigh schools criterion by criterion, and nearly every criterion lives on the program page.
Old pages get quoted as current. AI gave our nursing applicant 2018 and 2019 pass rates from schools' own old news releases.
Answers are unstable. Only 21–49% of named schools appear in all three runs of the same question.
How the study worked
Three fictional prospective students asked AI the questions real students ask, across five decision stages: explore, discover, qualify, compare and decide.
Psychology · Midwest
Emily
St. Charles, MO · 3.6 GPA · 26 ACT
“What does this degree lead to?”
Nursing · Southeast
Jasmine
Atlanta, GA · 3.8 GPA · 27 ACT
“Can I get into nursing as a freshman?”
Business · National
Ryan
Columbus, OH · 3.4 GPA · 30 ACT
“Where am I competitive?”
Questions
150 in total, 50 per persona. 33 are real Google search suggestions, word for word.
ChatGPT, Perplexity and Google AI Mode, set to each persona's home city. Every question ran three times.
Engines
Accuracy check
NCLEX pass rates AI stated, checked against the Georgia Board of Nursing's 2020–2023 results.
What students asked
Each major has its own core question. Nursing students ask how to get in and whether graduates pass. Psychology students ask what the degree leads to. Business students ask whether it's worth it and where they're competitive.
Google search suggestions by theme, from about 9,700 harvested
| Theme | Nursing | Psychology | Business |
|---|---|---|---|
| Best, top or ranked | 361 | 289 | 429 |
| Comparison ("vs", "or") | 142 | 472 | 452 |
| Admission (direct admit, accelerated) | 599 | - | - |
| Outcome proof (NCLEX, AACSB) | 239 | - | 232 |
| Worth it, jobs, salary | 32 | 110 | 150 |
| Undergraduate research | 0 | 94 | 4 |
Your program page content decides whether AI uses your words
FINDING 1
Across 32 schools, the share of a student’s checklist a school's pages answer correlates at 0.68 with how often AI cites the school's own site when it names the school. It has no positive link to being named at all.
| School (persona) | Checklist stated | Named in | Own site cited |
|---|---|---|---|
| Loyola University Chicago (Emily) | 7 of 8 | 28% | 77% |
| University of Missouri (Emily) | 5 of 8 | 44% | 66% |
| Georgia Southern University (Jasmine) | 8 of 9 | 12% | 58% |
| Miami University (Ryan) | 3 of 9 | 28% | 48% |
| Southeastern public university (Jasmine) | 7 of 9 | 24% | 23% |
| Elite business school A (Ryan) | 3 of 9 | 42% | 9% |
| Elite business school B (Ryan) | 1 of 9 | 30% | 4% |
| Highlight ranked private, Midwest (Emily) | 3 of 8 | 33% | 2% |
Famous schools are named on reputation. AI fills in their details from rankings, Poets&Quants and consultant blogs.
Specific pages get cited. Mizzou’s most-cited page is its psychology undergraduate research page; Miami’s are its business admission requirements and admitted-student statistics.
The wrong page can be the one cited. One university’s undergraduate answers came from its clinical psychology PhD handbook.
Comparisons hand the decision back to the student
FINDING 2
Of 270 comparison answers, 80% name no winner. They lay out the differences and end with “it depends on your goals.” The prize isn't being declared better. It’s having accurate specifics on each criterion the answer walks through.
Old pages get quoted as current
FINDING 3
We checked 164 NCLEX pass-rate claims about Georgia programs against the Georgia Board of Nursing's official 2020–2023 rates. Only 44 matched.
A private university
AI said
98%
from a 2019 news release
Board rate, 2023
94%
A large public university
AI said
100%
for a fall 2019 cohort
Board rate, 2023
84%
The pages AI cited were the schools’ own: a family and parents page, old news releases, a magazine brief, a BSN FAQ. A press release that was accurate in 2019 is still on the site, still indexed, and reads as fact with no expiration date.
A school's current pages don't have to be wrong for AI to be wrong.
Answers are unstable, so measure rates, not screenshots
FINDING 4
Ask the same question three times and you get three different short lists. Share of named schools that appeared in all three runs:
21%
ChatGPT
40%
Google AI Mode
49%
Perplexity
Any two engines agreed on only 36–41% of the schools they named. A single answer is a sample. Track the same panel of questions over time and report how often a school is named and cited.
Six changes for any academic program page
THE PLAYBOOK
01
State the facts students compare
Admission rules, outcomes, cost, research or clinical opportunities, in plain text on the page.
06
Name the program as students search
“BSN,” “psychology major,” in titles and headings.
02
Put a year on every outcome
“93% for the Class of 2023.” Undated figures become vague claims or old numbers.
03
Keep one current outcomes page
Update, date-label or redirect old news, parent pages and FAQs that quote outcomes.
04
Separate undergraduate from graduate
Be clear in titles, headings, URLs and cost pages that say “undergraduate.”
05
Answer the comparison criteria
What makes this program different on research, clinicals or internships, cost and outcomes.
By Program
Psychology
“What does this degree lead to?”
An undergraduate research page AI can quote
Where graduates go, with a year
Internships in counseling settings
BA vs BS, answered on the page
56% vs 32% own-site citation with a careers list vs without
Nursing
“Can I get in, and will I pass?”
NCLEX rates by year, three-plus years
Direct admission, yes or no
Named clinical partners
Audit every page quoting a pass rate
6 of 11 programs stated a dated NCLEX rate
Business
“Where am I competitive?”
Admitted-student profile, dated
Direct vs later admission explained
Placement and salary, with class year
Merit-aid eligibility a student can check
32% vs 14% own-site citation with an admitted profile vs without
Not supported yet: FAQ-style headings showed no link to being recommended, and structured data couldn't be measured reliably in this study.
Limits of this study
Test conditions, not the consumer apps. These are API answers under controlled settings. What a logged-in student sees in ChatGPT or Gemini may differ.
A small school sample. With 11–12 audited schools per persona, results are comparisons, not statistical models. The page-to-citation link is correlational.
Fictional personas. Only the 33 Google-suggestion questions are observed search behavior; the rest are modeled on it.
Directional search demand. Volumes come from autocomplete, not keyword tools.
One accuracy check. Accuracy was verified for Georgia nursing programs only, and board data ends at 2023.
Schools shown with challenges in visibility are anonymized; schools named are cited as examples of what works.
About this research
This study is part of ongoing Powers Digital research on how AI search treats higher education websites, alongside a review of AI crawler access across nearly 4,800 US colleges and a study of Google AI Overview citations. The next steps are a monthly re-run of the same 150 questions and a before-and-after test of these page changes on live program pages.
If you'd like to know how AI describes your programs, or to run this panel against your own schools, get in touch.
Sources
Powers Digital program search study, October 22–23, 2026: 1,350 AI answers, 5,088 school mentions, 8,735 citations, 7,230 extracted claims and 113 audited pages.
Georgia Board of Nursing, RN NCLEX Results 2020–2023 (rev. January 2024)