Query and expose feature usage share metrics

It would be nice* to expose the data usage of features HttpArchive tracks.
This can potentially be integrated with https://caniuse.com/ and https://www.chromestatus.com/features

*This would help in cases where you need to convince business to use feature(s) a competitor is leveraging even if not widely available.
or when Google announces the API and heavily market for it on events to be able to track where it’s implemented and not rely solely on use cases listed on case studies

Let’s say you are interested in adoption of PaymentRequestAPI and just my visiting either
CanIUse - PaymentRequestAPI or ChromeStatus - PaymentRequest you could see the adoption of it and maybe some urls depending on which category you pick (CMS or ECommerce).

I’m not that familiar with HttpArchive dataset but I’m getting used to it and find it enormously useful.
If I’m understanding right we could leverage Blink features and CSS Features.


I guess this is done by https://twitter.com/HTTPArchive/status/1085298896677597185.

I just didn’t follow up on this thread but managed to get the info needed quite easily https://twitter.com/matijagrcic/status/1064573932999753728

Uses the Blink and CSS features mentioned in the original post.

SELECT yyyymmdd, client, pct_urls, sample_urls
FROM `httparchive.blink_features.usage`
WHERE feature = 'IntersectionObserver_Constructor'
ORDER BY yyyymmdd DESC, client

Thanks for this question and the sample query!

A follow up question. When I run the query, I only see 20 result rows, with 10 sample_urls each. The pct_urls looks useful and believable. But the sample_urls don’t seem very representative of the top sites using the feature.

Is there a way to get a sorted list of the top sites using a feature, rather than what looks like a tiny random sampling of the long tail of sites that use it?

1 Like

To get more sample URLs per client you can query the blink_features.features table. The sample URLs are randomly generated from this table, so you can join these with the Alexa ranks and select the top ranked URLs.

That gives me exactly what I’m looking for. Thanks!

1 Like

Thanks for sharing :innocent:

1 Like