MBA-AEPI · Wave 1 · September 2026 · Public Release

The complete AI-visibility index for MBA programs.

Two modes of AI visibility: what ChatGPT already knows from training, and what it says once it can search live. Measured separately, scored honestly, and compared in one view. Wave 1: ChatGPT GPT-4o.

25 MBA programs 600 dual-mode prompts Memory + live retrieval Two scores, honestly separated

Download the methodology Get Your AEO Diagnostic

MBA-AEPI v1.9.0-FROZEN · Dual-mode CSPS protocol · ChatGPT GPT-4o · Collection date: September 2, 2026 · Public Tier 1 Release

New instrument This is Wave 1 of MBA-AEPI, a companion measurement to the AEO Index: M7 Business Schools. It uses a different methodology, a single-platform (ChatGPT) baseline, and a broader 25-program cohort. It does not replace the flagship M7 Index; see that page for the full 5-platform, memory-vs-retrieval measurement.

Programs tracked

Stanford 0.627

LSS leader

0.273

Median M₁ᴬᴳˢᴼ

Negative valence

25 Stable

ΔM₁ classification

1,200

Total captures

100%

Completion rate

Wave 1 What AI already says about MBA programs

The short version

Executive Summary

Everything this index measures, in ninety seconds. Every figure is drawn from the panels below, so the summary and the evidence cannot disagree.

0.627

Stanford GSB: #1 M₁ᴬᴳˢᴼ

Appears in 62.7% of all live-search responses across 600 prompts

Programs, all Stable

Zero Search-Amplified or Search-Suppressed at Wave 1 n=600

Negative valence

ChatGPT acts as a recommender, with no critical framing detected

0.63

MIT Sloan: top ECM

Most consistent mention rate across all four prompt strata

Finding Stanford GSB, Wharton, and Harvard Business School lead the AI-surface contest and are statistically tied: their Wilson 95% CIs overlap. Below Booth (rank 8, 0.442), visibility drops sharply. 17 of 25 programs score below 0.10, meaning ChatGPT mentions them in fewer than 1 in 10 responses. For those programs, AI visibility is an unaddressed competitive risk.

What this means M₁ᴬᴳˢᴼ is not a prestige index. It measures AI-surface salience, which diverges from US News rankings in measurable ways. Columbia Business School outscores Kellogg in LSS mode despite similar traditional rankings. MIT Sloan's ECM of 0.63 reflects cross-stratum robustness unavailable to schools with equal headline scores but narrow visibility profiles.

Act 1 Where programs stand

Primary metric

The M₁ᴬᴳˢᴼ headline index

M₁ᴬᴳˢᴼ = proportion of ChatGPT Live Search Surface responses in which each program is mentioned. n=600 prompts. Wilson 95% CI shown per program.

Finding The top-6 programs (Stanford, Wharton, HBS, MIT Sloan, Columbia, Kellogg) each exceed 0.57, a tier boundary roughly 3x above Tuck (0.280, rank 9). The gap between rank 8 (Booth, 0.442) and rank 9 (Tuck, 0.280) is the sharpest discontinuity in the distribution and likely reflects a M7 citation threshold in the training corpus.

M₁ᴺᴳᴹᴼ (TDK, training memory) M₁ᴬᴳˢᴼ (LSS, live search) ΔM₁ = LSS − TDK

1 Stanford GSB 0.627 +0.027

2 Wharton 0.625 −0.025

3 Harvard Business School 0.603 −0.002

4 MIT Sloan 0.597 +0.000

5 Columbia Business School 0.592 +0.002

6 Kellogg 0.573 −0.025

7 Haas 0.473 −0.022

8 Booth 0.442 +0.000

9 Tuck 0.280 −0.045

10 Yale SOM 0.275 +0.007

11 Stern 0.245 −0.005

12 Ross 0.188 −0.045

13 Fuqua 0.177 −0.013

14 Darden 0.122 +0.000

15 Anderson 0.117 −0.013

16 Tepper 0.087 −0.010

17 Marshall 0.073 −0.005

18 Foster 0.062 +0.002

19 McCombs 0.060 −0.028

20 Johnson (Cornell) 0.048 −0.022

21 Kenan-Flagler 0.045 +0.012

22 Mendoza / Olin 0.020 +0.000

24 Carey 0.015 +0.002

25 Smith (Maryland) 0.013 +0.002

Bars scale to the corpus maximum (0.627). Gray = TDK (training memory) · Blue = LSS (live search). ΔM₁ chip: green = search-amplified, gray = stable. Source: MBA-AEPI v1.9.0-FROZEN, September 2026.

Divergence analysis

Training data vs. live search: the ΔM₁ story

ΔM₁ = M₁ᴬᴳˢᴼ − M₁ᴺᴳᴹᴼ. Positive = live search amplifies the program beyond model memory. Option D dual-gate: |ΔM₁| ≥ 0.05 AND non-overlapping Wilson 95% CIs required for reclassification.

TDK Mode

Training Data Knowledge
No web search, model memory only
600 prompts collected

600Prompt Pairs
Each Run
Twice

CSPS Protocol

LSS Mode

Live Search Surface
Web search enabled, retrieval active
600 prompts collected

ΔM₁ Output

Amplified · Stable · Suppressed
All 25: Stable at Wave 1

Key finding All 25 programs are classified Stable. This is the expected result for a single engine at n=600: the Option D dual-gate (|ΔM₁| ≥ 0.05 AND non-overlapping CIs) is deliberately conservative. The notable exception is MIT Sloan's Factual stratum: M₁ᴬᴳˢᴼ = 0.200 vs. M₁ᴺᴳᴹᴼ = 0.067, a 3x within-stratum search amplification. Cross-engine expansion in Waves 2 to 5 will increase resolution and enable the first divergence classifications.

Program	M₁ᴬᴳˢᴼ LSS	M₁ᴺᴳᴹᴼ TDK	ΔM₁	\|ΔM₁\|	Class
Stanford GSB	0.627	0.600	+0.027	0.027	Stable
Wharton	0.625	0.650	−0.025	0.025	Stable
Harvard Business School	0.603	0.605	−0.002	0.002	Stable
MIT Sloan	0.597	0.597	+0.000	0.000	Stable
Columbia Business School	0.592	0.590	+0.002	0.002	Stable
Kellogg	0.573	0.598	−0.025	0.025	Stable
Haas	0.473	0.495	−0.022	0.022	Stable
Booth	0.442	0.442	+0.000	0.000	Stable
Tuck	0.280	0.325	−0.045	0.045	Stable
Yale SOM	0.275	0.268	+0.007	0.007	Stable
Stern	0.245	0.250	−0.005	0.005	Stable
Ross	0.188	0.233	−0.045	0.045	Stable
Fuqua	0.177	0.190	−0.013	0.013	Stable
Darden	0.122	0.122	+0.000	0.000	Stable
Anderson	0.117	0.130	−0.013	0.013	Stable

Top 15 programs shown. Full table of 25 in the Technical Appendix.

Act 2 How AI decides

Cross-stratum robustness

Engine Consistency Metric (ECM)

ECM = 1 minus CV(stratum M₁ᴬᴳˢᴼ scores). Range [0,1]. Higher = more consistent mention rate across Generic, Constraints, Persona, and Factual strata. Low ECM = prompt-fragile visibility.

Finding MIT Sloan leads ECM at 0.63. It is the only program that maintains meaningful visibility in Factual-stratum prompts (0.200 LSS vs. 0.050 baseline for most programs). McCombs and Kenan-Flagler have ECM ≥ 0.51 despite low headline scores, indicating consistent cross-stratum presence. Foster (ECM 0.00) and Johnson (ECM 0.06) are effectively single-stratum programs, nearly invisible outside their primary prompt type.

Program	ECM	Generic	Constraints	Persona	Factual	Profile
MIT Sloan	0.63	0.621	0.694	0.600	0.200	Stratum-robust
McCombs	0.52	0.088	0.056	0.017	0.050	Consistent low
Kenan-Flagler	0.51	0.050	0.044	0.017	0.083	Consistent low
Haas	0.49	0.517	0.533	0.508	0.050	Factual drop-off
Wharton	0.48	0.658	0.717	0.708	0.050	Factual drop-off
Booth	0.48	0.550	0.444	0.417	0.050	Generic-led
Harvard Business School	0.47	0.600	0.728	0.700	0.050	Factual drop-off
Columbia Business School	0.47	0.683	0.689	0.533	0.050	Factual drop-off
Anderson	0.45	0.117	0.183	0.050	0.050	Constraints-led
Kellogg	0.46	0.546	0.667	0.750	0.050	Persona-led
Stanford GSB	0.46	0.600	0.739	0.800	0.050	Persona-led
Marshall	0.42	0.100	0.089	0.008	0.050	Generic-led
Fuqua	0.41	0.104	0.228	0.308	0.050	Persona-led
Darden	0.39	0.154	0.044	0.208	0.050	Generic/Persona
Tuck	0.38	0.433	0.156	0.275	0.050	Generic-led

What this means A program with high M₁ but low ECM is competitively fragile: its AI visibility concentrates in one prompt type and collapses in others. Stanford GSB's Persona score of 0.800 (highest in the corpus) masks a Factual score of 0.050. Programs seeking to close the gap should invest in exactly the strata where they underperform.

Sentiment analysis

Valence: ChatGPT as recommender

Valence scored via keyword-pattern matching in ±200-character context windows around each entity mention. Positive = laudatory framing or list-position signal. Negative = critical or cautionary language.

Negative mentions, corpus-wide

Zero negative framing detected across all 1,200 responses and 25 programs

40.8%

Positive mention ratio

59.2% neutral. "Ranked," "leading," and "prestigious" language dominate

0.644

Foster: highest valence ratio

47 positive of 73 total mentions. Small n amplifies ratio.

0.640

Harvard Business School valence

464 positive of 725 total, the largest absolute positive-mention count

Finding ChatGPT operates as a pure recommender at Wave 1. It names programs, ranks them, and describes their attributes without critical or cautionary framing. This is structurally significant: the AI surface is not a neutral information channel, it is an advocacy channel. Programs that appear in AI responses are implicitly endorsed. Programs that do not appear are implicitly excluded.

Act 3 What this means for strategy

Framework

AEO vs. SEO: why AI visibility is a distinct discipline

Search Engine Optimization (SEO) targets indexing signals. AI Engine Optimization (AEO) operates on a fundamentally different mechanism: parametric memory compressed during training, sampled during generation.

The core distinction A program ranking #1 in Google for "top MBA programs" may still be structurally invisible to GPT-4o if its web presence did not generate sufficient co-citation mass in the training corpus. MBA-AEPI quantifies this gap. Columbia Business School (AI rank 5) and Kellogg (rank 6) surpass programs with comparable traditional prestige in specific prompt contexts. The AI surface rewards programs described frequently, positively, and in association with specific outcome signals in publicly accessible, high-authority web text.

Strategic interpretation

What your Stable classification means

Stable is not "safe." It is a statement that live-search and model-memory mention rates are statistically consistent within current measurement resolution. The strategic implication depends entirely on where in the index a program sits.

Classification	Profile	Strategic implication
High M₁ + Stable	Stanford, Wharton, HBS, MIT Sloan, Columbia, Kellogg	Deeply embedded in both TDK and LSS. Maintenance requires sustained output in AI-indexed channels.
Mid M₁ + Low ECM	Yale SOM, Tuck, Booth	Prompt-fragile: strong in generic queries, weak in constrained or factual ones. Targeted content strategy addresses this.
Low M₁ + Stable	Mendoza, Olin, Carey, Smith	Invisible in both modes. Must build AI citation mass from near zero. Longest investment horizon.

Act 4 What comes next

Multi-engine roadmap

Wave 2: Perplexity, coming September 10, 2026

Wave 2 launches September 10, 2026, replicating the full MBA-AEPI v1.9.0-FROZEN protocol against Perplexity AI and enabling the first cross-engine comparative index. All five platforms are on track to be live by September 30, 2026.

Wave 2 additions Perplexity's native inline citation architecture provides structured source metadata (domain, URL, date) unavailable in ChatGPT's non-citation responses. Wave 2 adds: cross-engine ΔM₁ (Spearman rank correlation), LLM-based valence scoring (replacing the keyword-window proxy), and a full citation-authority analysis by domain type.

Engine	Wave	Status	Architecture	Est. delivery
ChatGPT (GPT-4o)	Wave 1	✓ Complete	TDK + LSS (web search toggle)	September 2026
Perplexity AI	Wave 2	Planned	Native citation architecture	September 10, 2026
Claude (Anthropic)	Wave 3	Planned	TDK + web retrieval	By September 30, 2026
Microsoft Copilot	Wave 4	Planned	Bing-grounded retrieval	By September 30, 2026
Google Gemini	Wave 5	Planned	Google Search grounded	By September 30, 2026

Infrastructure How this is scored

Protocol

Methodology: CSPS v1.9.0-FROZEN

All MBA-AEPI waves use the Concurrent Synchronised Prompt Sequence (CSPS) protocol: TDK and LSS captures for each prompt pair, executed within the same session and 24-hour window.

600

Prompt pairs

240 Generic · 180 Constraints · 120 Persona · 60 Factual

1,200

Total captures

600 TDK + 600 LSS · 100% HTTP 200 · 0 failures

Target programs

All 25 detected ≥1 response · 0 zero-mention programs

Core scoring formulas

M₁ᴬᴳˢᴼ = mentions_LSS / 600: headline LSS mention rate

M₁ᴺᴳᴹᴼ = mentions_TDK / 600: TDK mention rate (model memory)

ΔM₁ = M₁ᴬᴳˢᴼ − M₁ᴺᴳᴹᴼ: divergence

ECM = max(0, 1 − σ/μ): consistency across 4 strata (σ=SD, μ=mean)

Wilson CI: (p̂ + z²/2n ± z·√(p̂(1−p̂)/n + z²/4n²)) / (1 + z²/n): α=0.05, z=1.96

ΔM₁ Option D: Search-Amplified if |ΔM₁| ≥ 0.05 AND CIs non-overlapping AND ΔM₁ > 0

Technical appendix

Data governance & complete entity table

Full 25-program table with disambiguation notes. All figures drawn directly from MBA-AEPI v1.9.0-FROZEN production run, September 2, 2026.

Rk	Program	M₁ᴬᴳˢᴼ	95% CI	M₁ᴺᴳᴹᴼ	ΔM₁	ECM	Valence	Mentions
1	Stanford GSB	0.627	[0.587, 0.664]	0.600	+0.027	0.46	0.62	736
2	Wharton	0.625	[0.586, 0.663]	0.650	−0.025	0.48	0.59	765
3	Harvard Business School	0.603	[0.564, 0.642]	0.605	−0.002	0.47	0.64	725
4	MIT Sloan	0.597	[0.557, 0.635]	0.597	+0.000	0.63	0.33	716
5	Columbia Business School	0.592	[0.552, 0.630]	0.590	+0.002	0.47	0.37	709
6	Kellogg	0.573	[0.533, 0.612]	0.598	−0.025	0.46	0.27	703
7	Haas	0.473	[0.434, 0.513]	0.495	−0.022	0.49	0.33	581
8	Booth	0.442	[0.402, 0.482]	0.442	+0.000	0.48	0.32	530
9	Tuck	0.280	[0.246, 0.317]	0.325	−0.045	0.38	0.32	363
10	Yale SOM	0.275	[0.241, 0.312]	0.268	+0.007	0.22	0.46	326
11	Stern	0.245	[0.212, 0.281]	0.250	−0.005	0.43	0.36	297
12	Ross	0.188	[0.159, 0.222]	0.233	−0.045	0.32	0.19	253
13	Fuqua	0.177	[0.148, 0.209]	0.190	−0.013	0.41	0.25	220
14	Darden	0.122	[0.098, 0.150]	0.122	+0.000	0.39	0.32	146
15	Anderson	0.117	[0.093, 0.145]	0.130	−0.013	0.45	0.24	148
16	Tepper	0.087	[0.067, 0.112]	0.097	−0.010	0.12	0.32	110
17	Marshall	0.073	[0.055, 0.097]	0.078	−0.005	0.42	0.33	91
18	Foster	0.062	[0.045, 0.084]	0.060	+0.002	0.00	0.64	73
19	McCombs	0.060	[0.044, 0.082]	0.088	−0.028	0.52	0.37	89
20	Johnson (Cornell)	0.048	[0.034, 0.069]	0.070	−0.022	0.06	0.39	71
21	Kenan-Flagler	0.045	[0.031, 0.065]	0.033	+0.012	0.51	0.28	47
22	Mendoza	0.020	[0.011, 0.035]	0.018	+0.002	0.00	0.26	23
22	Olin (WashU)	0.020	[0.011, 0.035]	0.020	+0.000	0.00	0.08	24
24	Carey (Johns Hopkins)	0.015	[0.008, 0.028]	0.013	+0.002	0.00	0.53	17
25	Smith (Maryland)	0.013	[0.007, 0.026]	0.012	+0.002	0.00	0.33	15

Source: MBA-AEPI Wave 1 production run · ChatGPT GPT-4o · September 2, 2026 · v1.9.0-FROZEN · 1,200 captures · 100% completion · 0 anomalies

FAQ

Common questions about MBA-AEPI

What is MBA-AEPI?

MBA-AEPI (MBA AI Engine Perception Index) measures how often ChatGPT names each of 25 MBA programs across two modes: what the model already knows from training, and what it says once it can search the live web. Wave 1 covers ChatGPT (GPT-4o) only, with 1,200 responses across 600 dual-mode prompt pairs.

How is MBA-AEPI different from Arcalea's M7 AEO Index?

They are separate instruments. The M7 Index tracks M7 business schools across five AI platforms with a five-signal composite score and a memory-versus-retrieval quadrant. MBA-AEPI Wave 1 tracks a broader 25-program cohort on ChatGPT only, using a training-memory-versus-live-search comparison and a different scoring model built around Wilson confidence intervals. Neither replaces the other; MBA-AEPI is a newer, narrower baseline that will expand across engines in later waves.

What does a Stable classification mean?

Stable means a program's live-search and training-memory mention rates are not reliably different once measurement error is accounted for. It is the default result at Wave 1: with a single engine and n=600, the Option D rule (a confidence-interval gap of at least 0.05, with non-overlapping intervals) is deliberately conservative, so most programs land here.

Why does Wave 1 use ChatGPT only?

MBA-AEPI is built as a longitudinal instrument, adding one engine at a time so each wave's methodology is fully validated before the next is layered on. ChatGPT (GPT-4o) is the Wave 1 baseline. Perplexity launches Wave 2 on September 10, 2026, followed by Claude, Copilot, and Gemini, all targeted to be live by September 30, 2026.

How is AI visibility different from a traditional MBA ranking?

A traditional ranking (US News, Financial Times) scores programs on inputs like selectivity, salary outcomes, and survey results. MBA-AEPI scores something else: how often and how consistently ChatGPT actually names a program when someone asks it for guidance. A program can rank well in traditional guides and still be nearly invisible in AI answers, or the reverse, since AI visibility depends on how much citable, co-referenced content about a program exists in the model's training data and on the live web.

What is the Engine Consistency Metric (ECM)?

ECM measures how evenly a program is mentioned across four prompt types: Generic, Constraints, Persona, and Factual. A program with a high headline score but low ECM is winning in one context and nearly absent in others, which makes its AI visibility fragile rather than durable.