Hi CrUX users,
This is your monthly announcement that the latest dataset has been published to BigQuery.
The 202205 (May 2022) dataset is now available and it covers 11,024,795 origins, an increase of 28.2% over last month! This increase in origin coverage is thanks to a new feature that we're rolling out called optional dimensions. More on that later, but first here’s a look at origins' Core Web Vitals performance this month:
The percentage of origins having good LCP experiences increased this month by 2.4%, thanks in large part to an improvement in the performance of Chrome on Android. This was a major contributing factor to the percentage of origins having good Core Web Vitals increasing by 2.1%.
Until now, records in CrUX have had required dimensions like effective connection type and form factor fields. This has historically been in place as a precaution for data quality assurance, but recently we've taken a closer look and decided that we can confidently remove this requirement while still ensuring high data quality. The change is to make these dimensions optional.
Therefore, CrUX users may see a value of NULL indicating that we don’t have sufficient data for a more specific form factor and/or effective connection type. In other words, you can think of NULL as inclusive of all possible values. This makes it easier for pages and origins to be included in the CrUX dataset, and we're excited about the expansion in coverage that it provides for site owners, because having some coarser data is better than having none.
Unfortunately this month's release coincides with an unrelated issue with our data pipeline. Each month, there is a natural amount of churn in origins either dropping out of the dataset or being newly added to it. This month though, we were unable to add any new origins that have never been seen before. So how were we able to increase coverage by 28.2%? We reintroduced any origins seen in the last 6 months whose May 2022 data is now considered sufficient according to the new optional dimensions criteria. We hope that in next month's release we're able to fully roll out this change and realize an even greater expansion in coverage.
As a side-effect of this issue, the experimental.popularity.rank field has some gaps in it. For example, there are only 904 origins ranked in the top 1,000 because we had to filter out 96 of those origins from the pre-release dataset. This is unique to the 202205 release and should be resolved next month.
If you have any questions about CrUX, feel free to reach out to us on any of these channels:
Rick & Johannes