How accurate is Similarweb data for subdomains?

Even though Similarweb manages to do reasonably well when it comes to the Spearmans correlation with the actual website traffic, in our personal experience the accuracy reduces considerably when using similarweb for subdomains. As Rand Fishkin from Moz points out in a test he ran comparing the major traffic estimation tools "I wouldn’t feel confident using any of these numbers to predict actual traffic, but I would recommend using a combination of SimilarWeb, Compete, and (when available) Quantcast to get a rough sense for relative traffic between sites".
Heres the link to his full test https://moz.com/rand/traffic-prediction-accuracy-12-metrics-compete-alexa-similarweb/