Virgin Speed Tests

I am on Virgin 200Meg. When doing various speed tests I find that speedtest.net gives much higher results than all the other sites, sometimes in excess of 200Meg download which seems wrong. Then I found this on the Virgin site I quote "There are many speed test sites available, however we recommend using speedtest.net as we�ve worked with the site to offer reliable tests to our customers."
Anyone else noticed this?
Seems they are fiddling it?
Can anyone shed any light?

Virgin Media generally over provisions so a good speed test on a 200 Mbps connection will be a little above 200 Mbps e.g. 204 or 205 Mbps.

The sampling methodology used by speedtest.net means the result is closer to a top 25% or 20% figure during the test, compared to for example us showing a median (we include a burst which is set at top 15% figure).

Another issue is that Virgin Media have ookla testers on their own network and as speedtest.net picks the server with the lowest latency they will favour on net providers testers.

Presumably the placing of speed testers at Virgin network sites is a good idea as this allows users to test the Virgin network aspects of speed and thus see if there is anything wrong with that; overloading of local network for example. Admittedly it excludes testing of interconnects and transit within the the wider internet, which may occasionally be an issue, but speedtest users can select a test site further away to include this. With regard to fiddling tests, I can't see how this could be done and I too have noticed that Virgin seem to overprovision speeds a little these days. (Unlike DSL Virgin speeds are set in the network equipment and are not determined by distance from exchange/headend). To make certain, download a big file, remembering to multiply file size in MBytes by 8,388,608 (1024 x 1024 8 bit bytes per Megabyte) by seconds to download to get the bitrate.

Edited by deleted (Tue 08-Aug-17 18:42:49)

Testing using payloads from a providers network is useful, but if a provider uses that to say there are no problems then it can be misleading if the provider knows some parts of it core network to different locations are congested.

The methodology that speedtest.net uses can often over report - I have had speeds of 30 to 40% above my line capabilities. Just be cautious of the results from there.

This result was achieved on a 2Mbps link! http://www.speedtest.net/result/384967663.png

Good grief! I will be cautious

I'd certainly agree that any claims there are no problems by a provider who knows there are is reprehensible. In my experience problems in the local network are much more common than in the core/backbone so eliminating this is a useful first step. Then using more remote test sites if there are still issues and the local is ok.

That 2M result is amazing and certainly appears to show that speed test results can mislead. Not quite the same as fiddling the results though.

For what it's worth, I run Speedtest and Thinkbroadband tests weekly or so, and find they track each other remarkably closely...my results from a couple of minutes ago:

Speedtest: 74.93Mbps down/17.27 Mbps up

Thinkbroadband: 74.9Mbps down/17.0Mbps up

I was an extreme oddity, however I did have quite a few when on ADSLx and 40M VDSL that were 20, 30, 40% over what could be achieved and often they were when using non-UK servers too.

Mine showing approx an 80Mbs download difference using the Thinkbroadband burst and Speedtest beta using optimal server. Run a few moments ago.