crissyfield/crux-dumps

As detailed in this research paper, the Chrome UX Report website data is significantly more accurate than other website ranking lists, such as the now defunct Alexa's Top-Million or the Tranco List:

Using a set of metrics from Cloudflare that estimate page loads and unique visitors, we find Google Chrome's recently released CrUX dataset captures the unordered set of most popular websites significantly more accurately than other top lists, with correlations inline with the differences we see amongst multiple measures of popularity derived from the same Cloudflare data. No other top list enters this range. This, paired with the internal consistency of Chrome metrics, suggests that Chrome does not simply use a metric more similar to Cloudflare’s, but rather that their data is more accurate.

Dumps in this repository are automatically generated by exporting the origin and rank columns from the CrUX dataset via BigQuery, grouping URLs by rank, and storing the grouped URLs as XZ compressed archives.

Leave a Comment