The index as files: every business, every month

One line of DuckDB

DuckDB reads the CSV straight from the URL; so does pandas, Polars, R and a spreadsheet.

SELECT city, count(*) AS sites, round(100.0 * avg(channel IN ('first_party', 'saas')), 1) AS direct_pct, round(100.0 * avg(blocked), 1) AS blocked_pct
FROM read_csv('https://directindex.org/data/2026-10/all.csv')
WHERE reachable IS NOT NULL GROUP BY city ORDER BY direct_pct DESC;

Same data, live: /api/city/{slug} (the index as JSON), /city/{slug}/{month}?format=json (a monthly snapshot), /api/city/{slug}/changes (what changed), OpenAPI, MCP, the quarterly edition.

October 2026: the month so far

Written 2026-10-14 23:22 UTC from the latest scan of each site; regenerated daily until the month ends. all.csv, every business across every index, 11.6 MB · index.json.

Every scan ever made, as a table

Behind the monthly files sits the history: every scan row, including the ones these files summarise, lands in an Apache Iceberg table (direct_index.scans) in R2 Data Catalog, on the bucket direct-index-history. One row per scan with the score, its components, the menu format, the channel, the vendor, the builder, the crawler findings and the full result as JSON. Researchers, journalists and agent builders can query it with any Iceberg client (DuckDB, Spark, PyIceberg, Trino). Access is on request: write to hello@directindex.org with what you want to look at, and you get a read token.

ODbL, with attribution

Business listings come from OpenStreetMap contributors under the Open Database Licence (ODbL 1.0). The scan results, scores and findings are published by The Direct Index under the same licence. You may copy, share, adapt and build on the data, for any purpose, as long as you attribute "OpenStreetMap contributors" and "The Direct Index (directindex.org)" and share any adapted database under the same licence. Numbers carry the date they were measured; cite the month.