MapTheNet

Open data

Download the data.

The rankings and the link graph behind them, free to download under CC BY-SA 4.0.

Domains

14,444,633

Links

22,545,058

Snapshot

2026-10-30

map.png 34.9 MB domains.csv.gz 253.7 MB links.csv.gz 172.3 MB README.txt 2 KB manifest.json 886 B

Using it

The dataset is under Creative Commons Attribution-ShareAlike 4.0. The software that makes it is under the GNU Affero General Public License, version 3.

You can

  • Use it commercially, including in a paid product
  • Share it, changed or unchanged
  • Build datasets, models and visualisations from it

You need to

  • Credit MapTheNet and link to this site
  • Say which snapshot you used. The index changes with every build
  • Publish anything derived from it under the same licence (the ShareAlike condition)

Please don't

  • Present the rank as traffic, revenue or quality. It only measures links
  • Try to identify people from the link words. They're published only as totals across many sites
  • Republish domains whose owners asked to be removed. The exclusion list is in the download
  • Imply that we endorse your work

An attribution that covers the licence:

MapTheNet (https://www.mapthenet.org), snapshot 2026-10-30, CC BY-SA 4.0

What's inside

domains.csv.gz has one row per domain: rank, score, country, TLD, referring domain count, outbound link count, community id and crawl status.

links.csv.gz is the graph itself: which domain links to which, whether the link is nofollow or on every page, and when we first saw it.

The domain file has every ranked domain. The link file has the links between the top 2,000,000.

The full dataset. The downloads stop at the top 2,000,000 domains. The full export is several hundred GB, and this server has no room to host it. We are looking into ways to publish the full dataset. Until then, for research, journalism or other projects that need all of it, email [email protected] with who you are, what you plan to use it for, and how you can receive a transfer of that size (for example, storage we can upload to). We look at each request individually and can't take all of them. The full export leaves out the same fields as the download.

CSV has no row limit, but Excel opens only the first 1,048,576 rows. For larger files, use a database, pandas or DuckDB.

What's left out

  • each domain's trust, link growth and network concentration scores (the method is published, the individual scores aren't)
  • when each site is due for a crawl, and robots.txt fingerprints
  • page content, URLs, IP addresses and WHOIS records, which we never collect

The ranking method is fully published. What we leave out is each domain's score on the individual anti-spam checks, because link farms could use those to work out how to get around them.

What the numbers mean

Rank shows how well connected a domain is through links. It isn't traffic, popularity, revenue, quality or trustworthiness. The methodology page explains how it's calculated and what it can't see.

The download is a snapshot of the last rebuild and doesn't change in between.