Collaborate

The Alexa data we use to calculate website popularity has 2x issues:

1. It's increasingly inaccurate: its data is desktop only (based on downloads of the Alexa toolbar), whereas traffic has substantially shifted to mobile
2. It can't handle subdomains, e.g. Groups.Google.com, or News.Yahoo.com. (Or some Baidu forums). As such, these sites throw all our visualisations of top sites, top content, and top influencers out -- researchers have to exclude these channels from our dataset before doing analysis. (We've been saying this is an issue for years, but never built in a manual fix.)

Solution: we need to find a better estimate of web track

Possible solution: SimilarWeb API
https://www.similarweb.com/corp/developer/

"We leverage hundreds of sources which we categorize into 4 distinct groups:
1. Global Panel Data from hundreds of millions of desktop/mobile devices
2. Global ISP Data from partners with millions of subscribers
3. Public Data Sources from over a billion sites and app pages every month
4. Direct Measurement Data from hundreds of thousands of sites and apps"

This level of multi-source triangulation - and coverage of mobile app behaviour - suggests it's quite likely to be more accurate than Alexa. We should request comparative data and get pricing, to consider swapping integration.