Great data, and definitely more accurate than the raw number of domains check, but I think it falls short quite frequently… Just looking at a handful of small sites, you quickly come across 1st party requests that don’t fit the pattern:
- cnn.com uses cdn.turner.com, clearly a 1st party
- walmart.com uses *.walmartimages.com, again clearly a 1st party
- facebook.com uses akamaihd.com, which is actually a 1st party (also uses fbcdn.com)
- twitter.com uses twimg.com, clearly a 1st party
I think this is a step ahead, and it’s also good to see the number of requests, not just number of domains, that are fetched from the same top level domain vs not, but I think it still falls short when identifying 3rd party content.