AI Visibility Index
The Plexa Index found that AI names no business from the directory in 62.3% of answers about local services in Mexico.
Measured across 922 real AI answers (ChatGPT and Google AI Overviews) between June 14, 2026 and July 14, 2026. Figure from the First Edition (corrected), the one the DOI fixes. Methodology and sample, below.
Current measurement · in the most recent edition (589 answers, not yet under a DOI) silence rises to 65.2%: purging directory phantoms reveals more real silence.
Finding 1 · the silence
In 62.3% of AI answers, AI names no business from the directory.
Someone asks for a local service and it answers with no concrete name, or with businesses not in the registry. Most answers come up empty.
Finding 2 · the concentration
When it does name someone, the 3 most-cited businesses capture 40.4% of their category's mentions.
Visibility is not spread out: it concentrates in a handful of names per vertical and city. Everyone else competes for the leftovers.
Finding 3 · the low ceiling
Nobody dominates: the most visible business barely reaches 30.8 out of 100 on the AI Visibility Score.
Of the 424 businesses AI named at least once, 379 sit below 10 points. The field is wide open for whoever structures first.
The AI Visibility Score (0-100)
Per business = the fraction of its category's AI answers it appears in, ×100. How the 424 businesses AI named break down:
| Score 1 – 10 | 379 businesses |
| Score 11 – 25 | 44 businesses |
| Score over 25 | 1 businesses |
By vertical
| Vertical | Answers | Businesses cited | No names |
|---|---|---|---|
| Dentists | 370 | 144 | 54.6% |
| Doctors | 152 | 52 | 76.3% |
| Restaurants | 136 | 60 | 69.1% |
| Tours | 96 | 44 | 65.6% |
| Real estate | 96 | 49 | 68.8% |
| Wellness | 24 | 22 | 45.8% |
| Hotels | 24 | 43 | 41.7% |
| Cars | 24 | 29 | 50% |
A business can be named in more than one vertical; that is why the sum of the businesses-cited column (443) exceeds the total of unique businesses (424).
By city
| City | Answers | Combos | No names |
|---|---|---|---|
| Cabo San Lucas | 464 | 8 | 60.1% |
| Ciudad de México | 136 | 5 | 68.4% |
| Guadalajara | 82 | 2 | 68.3% |
| Tijuana | 60 | 2 | 45% |
| Monterrey | 36 | 2 | 69.4% |
| Medellín | 24 | 1 | 58.3% |
| Playa del Carmen | 24 | 1 | 75% |
| Puerto Vallarta | 24 | 1 | 66.7% |
| Mazatlán | 24 | 1 | 66.7% |
| Vicente Guerrero | 24 | 1 | 58.3% |
| Zapopan | 24 | 1 | 66.7% |
Methodology
We monitor the answers of 25 combos (vertical × city) across 8 verticals and 11 cities, running real customer queries against ChatGPT (OpenAI) and Google AI (Google). Within the window we measured 922 answers and recorded 1,704 mentions of 424 distinct businesses from the directory. Each market was measured in one or more runs across the 30-day window; the number of answers per market reflects how many times it was measured, not a difference in method.
The sample is concentrated: 50.3% of the answers (464 of 922) come from a single metropolitan area, Cabo San Lucas, which weights the global aggregates toward that market.
The «silence» is measured at the answer level (answers that name no business from the directory), not as a «% of invisible businesses»: the monitoring sees whom AI cites, not the full universe, so we do not invent a denominator. Google AI is queried as Google AI Overviews through a monitoring API; the measurement is faithful to each engine's behavior, not an exact replica of every individual session. House honesty: in this wave, Plexa itself was cited in 0.1% of answers (1). We are entrants; we measure it anyway.
Rankings in this index are not for sale: no business can pay to appear or to improve its position. Some engines already run paid ads; this index measures organic mentions only.
Edición 1 (corregida · matcher fix 17-jul-2026) · basado en 922 respuestas de IA de 25 combos (8 giros × 11 ciudades) medidas entre 2026-06-14 y 2026-07-14, en ChatGPT (OpenAI) y Google AI. Silencio recomputado 71.6% → 62.3%.
How to cite
This study is published with a DOI on Zenodo, under a Creative Commons BY 4.0 license.
Plexa; Cortes, C. (2026). AI Visibility Index, First Edition (corrected). Zenodo. DOI: 10.5281/zenodo.21433409
This is the corrected First Edition (v2): a matcher correction on 2026-07-17 recomputed the silence from 71.6% to 62.3% over the same window and the same 922 answers (methodology). The original version remains published, immutable, as a superseded historical snapshot: DOI 10.5281/zenodo.21364735.
Download the study (PDF) →