The problem with this is that Fast.com is owned by Netflix, which is such a significant consumer of bandwidth that it inclines ISPs to shape traffic to fudge the numbers. Essentially, this kind of speed test only really tells you what the maximum bandwidth your ISP is able to provide you is, rather than what they're really providing. Even then, the test result with Fast.com could be subject to traffic shaping or slowdown at the remote end, so what is the result really telling you at all?
In fairness, this would be true of any popular speed test, and I'm not sure of an obvious solution. Any kind of tunneling or onion routing would render the test pointless anyway, so I'm not sure of the extent to which you can ever really trust a speed test of this nature. The only really useful data for this kind of analysis is likely to come from a router with sufficiently sophisticated software that it can report on transfer rates over time (which many do), and even then, that doesn't account for scenarios where the remote connection is the bottleneck.
In fairness, this would be true of any popular speed test, and I'm not sure of an obvious solution. Any kind of tunneling or onion routing would render the test pointless anyway, so I'm not sure of the extent to which you can ever really trust a speed test of this nature. The only really useful data for this kind of analysis is likely to come from a router with sufficiently sophisticated software that it can report on transfer rates over time (which many do), and even then, that doesn't account for scenarios where the remote connection is the bottleneck.