Interesting! I always thought, this kind of data was collected from a plethora of borderline shady browser plugins (SEO plugins amongst them, ironically), toolbars and alt-browsers.
This is also, why I think the data must be heavily biased. For example the HN demographic must be severely underrepresented, because most of us are careful what we install.
B2B keywords typically have 2-5x higher search volume/month than the popular tools estimate, for exactly the reason you provided. More sophisticated searchers aren't installing random anti-virus, VPNs, and plugins the SEO tools rely on to calculate keyword search volume.
This is also, why I think the data must be heavily biased. For example the HN demographic must be severely underrepresented, because most of us are careful what we install.