Alexa rank is one of the most popular metrics that characterizes a website’s traffic both globally and locally (by country). The lower the Alexa rank, the more traffic a website has. Millions of internet users have Alexa toolbar installed – and if you have your toolbar installed, Alexa can gather information from your browsing habit as well in order to calculate Alexa rank of websites.
Of course, as every tool, Alexa has its limitations and potential errors in its calculation as well. Its error in calculation mainly comes from the fact that they only have access to someone’s browsing habit only if the Alexa toolbar is installed and than the data is weighted by the number of country’s internet users which might falsify the data a bit. Obviously, if you have Alexa tool installed, you already belong to the group of more advanced internet users. Thus, websites dealing with advanced internet science topics like search engine optimization or link directories are always overrated by Alexa. This can mean a even a hundredfold difference in traffic of two websites possessing similar Alexa rank.
Another widely used tool for traffic calculation is SimilarWeb. They have more channels of data collection: from millions of monitored devices, from local internet providers, form their own crawlers and they also have many websites and apps that are directly connected to their system. SimilarWeb’s ranking is the same as Alexa’s, so the lower the rank, the higher the website’s traffic.
As the ranking logic of the two tools is the same, we wanted to see the correlation between them. We have analyzed 500 domains that possess Alexa and SimilarWeb values within 1 000 000. Obviously, if the two metrics are perfect, they should yield the same rank on the websites.
However, as you see from the Figure, there is only a slight correlation observed between these two rank values. There is a large group of low Alexa, but high SimilarWeb ranking websites. This group indicates the biased nature of Alexa rank as discussed above.
So let us see what happens if we go to higher than 1 000 000 ranking – where, of course, traffic estimation is more inaccurate.
As you see, above 1 million the two ranks have no correlation at all. This is because at Alexa or SimilarWeb rank above 1 million, the small number of visitors do not make accurate traffic estimation possible.