All three are testing slightly different things, speedtest.net tends towards showing a burst speed (we show avg and burst but our burst is calculated slightly differently). What bt do, no idea but their test can bypass a chunk of internet too, and relies on java to add to the difference.

In reply to a post by MrSaffron:

What bt do, no idea but their test can bypass a chunk of internet too, and relies on java to add to the difference.

BTW speed tester no longer uses Java.

In reply to a post by spranglebolt:

It's interesting that, if you run the bt speedtest, tbb, and speedtest.net in quick succession (how sad is that?) you can get three totally different results which I don't believe can be just the result of a couple of minutes time delay.

Why is that? Which do you believe?

Probably because each of the testers is cherry-picking different bits of the data to disregard. For example, speedtest.net discard the fastest 10% and slowest 30% of the samples in the test (all explained at https://support.speedtest.net/entries/20862782-how-d... ).

Thanks, and there are often large differences in ping as well, IMHO too great to be accounted for by the couple of minutes time difference - speedtest usually being well in excess of tbb.