We've improved our ranking algorithm by adding traffic estimates.
Before this update, we calculated rank for domains in our database based on harmonic centrality and page rank from the Common Crawl dataset. Centrality and page rank are based on the links between domains. They are a good indication of the connectedness of a domain but are not the best indicator of a domain's popularity. As a result, some of the top domains under the old algorithm were a bit of a headscratcher.
In an ideal world, rank would be based on merchant sales volume but this information is not available. Merchants just aren't in the habit of publishing the data. But we can approximate sales based on how much traffic a domain gets. While there isn't a reliable source of traffic estimates that applies to all stores, we were able to find and integrate traffic estimates that result in a noticeable improvement to our ranking algorithm.
We'll continue to refine our ranking formula so expect things to keep improving over time.