# Agentic commerce in Mexico 2026 · open data

The figures behind the study's eight charts, published as files so anyone can cite, reproduce or contrast them.

- **Study:** https://www.edgebound.ai/en/labs/agentic-commerce-mexico-2026
- **Method:** https://www.edgebound.ai/en/labs/agentic-commerce-mexico-2026#method
- **Published:** September 11, 2026 · **Fieldwork:** September 7, 2026
- **License:** [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)

## What is in the package

| File | What it holds |
| :--- | :--- |
| `agentic-commerce-mexico-2026.json` | The eight tables with their titles, subtitles and study metadata |
| `agentic-commerce-mexico-2026.csv` | The same figures in long format, one row per cell |
| `README.md` | This file |

Both files are extracted from the tables on the page, so file and study cannot diverge.

## Disclosure of interest

Study designed, executed and funded by Edgebound Labs. No evaluated retailer sponsored it, participated in its design, or had access to the results prior to publication. Edgebound Labs sells services covering the same areas the instrument evaluates. The measurement was not commissioned or endorsed by CANACINTRA or by any other chamber or trade association.

## Suggested citation

> Edgebound Labs (2026). *Agentic commerce in Mexico 2026: a study of 14 retailers.* https://www.edgebound.ai/en/labs/agentic-commerce-mexico-2026

The license asks you to credit Edgebound Labs and link to the study. It does not ask for permission.

## Field dictionary

### CSV, long format

One row per table cell. Convenient for pivoting in a spreadsheet or loading with pandas.

| Column | Type | What it is |
| :--- | :--- | :--- |
| `figure` | integer 1 to 8 | Figure number in the study |
| `figure_title` | text | Figure title, as published |
| `row` | integer | Row index within that figure, starting at 1 |
| `column` | text | Name of the column the value belongs to |
| `value` | text | The datum. Kept as text because proportions ("11 of 11"), counts and decimal scores share the same column |

### JSON

Metadata at the root level and the eight tables under `tables`.

| Field | What it is |
| :--- | :--- |
| `name` | Name of the dataset |
| `license`, `license_name` | License URL and name |
| `attribution` | Attribution required by the license |
| `disclosure` | Disclosure of interest, the same one the study publishes |
| `methodology_url` | Direct anchor to the method section |
| `creator` | Who produced the measurement |
| `source_page` | Page the figures were extracted from |
| `inLanguage` | Language of this edition |
| `datePublished` | Publication date of the study |
| `measurementDate` | Fieldwork date, a single run |
| `spatialCoverage` | Geographic coverage |
| `sample` | Sample breakdown: measured, scored, blocked at the edge, and insufficient evidence |
| `instrument` | Instrument used to measure |
| `checks_per_site` | Checks in the instrument |
| `areas_measured` | Maturity areas the checks are grouped into |
| `note` | Scope and limits of the file |
| `extracted_at` | When the figures were extracted from the page |
| `tables[]` | `figure`, `title`, `subtitle`, `columns` and `rows`. Each element of `rows` is an object whose keys are those in `columns` |


## What the files do not include

Store identities are not included. The study does not name brands next to defects: it publishes only aggregates and ranges by category. That rule applies to these files too.

They also exclude per-site raw data and per-check evidence. What is published is what backs the eight charts.

## Reproducing it yourself

The instrument and the full method are public on the study page. Any third party can repeat the measurement with their own team or with the vendor of their choice, and contrast the results.