[evanelias]

fwiw I have seen situations where their numbers were completely detached from reality, in absolute terms as well as relative comparisons over time.

They're especially inaccurate for any product with mobile apps, since they have zero insight into that traffic. Nor do they properly extrapolate between the massively different traffic patterns of users with apps (which tends to be a lot stickier) vs web traffic.

Several times, people citing SimilarWeb numbers have argued with me about things where I had first-hand direct knowledge of the correct numbers (e.g., numbers involving a company I worked for). It's frustrating. These people sided with SimilarWeb because it's some source they can cite, nevermind the fact that it was complete garbage in these cases.