Why do BT and TBB speed testers differ so much?

SInce upgrading to FTTC in January I have been puzzled by the difference in download speed as measured by the BT speed tester and the ThinkBroadband one (and others).

My IP download profile is 29.23 Mbps, upload 2Mbps.
BT speed tester typically measures 20-24 Mbps (it went down to 6 for a while but that is a different issue). For the first week or so after the upgrade, the TBB tester agreed with this. Some other testers (eg broadbandspeedchecker.co.uk) still do.

Now, however, the TBB speed tester typically measures about 40% of this ie 8-10 Mbps.

When I had ADSL, the BT and TBB checkers used to agree pretty well on 2.1 Mbps.
What is the experience of other new fibre users?
Are these checkers measuring the same thing?

With FTTC the capacity out of the providers network is more critical

The BT speedtester avoids that bit of the network.

Question is which checker is more in line with your actual day to day activity?

That is clearly a rhetorical question... In terms of what I see day-to-day (browsing, downloads, streaming, latency) I haven't noticed even a 5-fold improvement never mind a 10-fold one. I suppose perceived performance is not directly related to max speed though... also, maybe I'm now being limited by the servers rather than the client.

When you say the "provider's network" do you mean ISP or the providers of the servers I am connecting to?

Things I've seen that cause limitations in speed on FTTC:

- Using wireless instead of wired
- A congested wireless channel, through neighbours using same channel
- Wired ethernet falling back to a 10Mbps connection, instead of 100Mbps
- Plugins in Firefox
- Traffic Management at ISP

The starting point is to know what speed you *ought* to be getting. What IP Profile are you - as shown by the BT speedtester, but not the Beta version?

I usually find the speedtest.net tester most consistent, when using the Newbury server.

Being a muppet and letting win7 make your network connection be a "public" rather than "home" network. Shaves a good 10% off TBB, which i found yesterday when i had FTTC installed.

I find the BT, TBB Java and TBB Flash correspond reasonably well - allowing for differences in units kbps, Mbps and1024 divisors (or not). They track very well and if one is down by a significant amount the others will be too.

Have just run a comparison - I don't know if Saturday morning is a quiet time out there in Internetland, but here are the figures:

TBB speed test - 7.6M download, 1.6M upload
BT speedtester - 42.0M download, 2.1M upload (sic)
speedtest.net - 34.7M download, 1.9M upload

Given that the theoretical maximum for my current line profile is 31.6M down, 2.0M up, how can the figures from BT and speedtest.net be accounted for? Maybe BT is different, but why is TBB so much lower than speedtest.net? Is it server-dependent? (I did get a reading of 19.5M out of it last week - but only briefly).