Skip Navigation

Confusion reigns over broadband speeds
Monday 20 January 2014 11:02:05 by Andrew Ferguson

The 'woe is me' attitude that pervades so much broadband coverage is nicely summarised by some of the coverage for the Playstation Now game streaming service, that has been announced but is yet to launch but will basically be the Sony competition to OnLive.

It seems some people have latched onto the Netflix speed index results, which suggest that for example Virgin Media has a speed of 2.99 Mbps when viewing Netflix streams. There is a school of thought that this is reflecting the true position, rather than what the politicians want us to believe is the real world speed. The reason why Netflix speed index results should NOT be used as real world indicators is shown when even Google Fiber is listed as a 3.69 Mbps average speed.

The reality on broadband speeds is that many millions in the UK can get world leading speeds, and while we are still in the position that our own speed test puts Wales down the bottom of the league table with a median speed of 6.6 Mbps, areas like Northern Ireland that has had wider coverage of faster broadband speeds for some time is showing a median of 13.9 Mbps. So yes some people are not going to have the pre-requisite 5 Mbps minimum speed needed to get a reasonable Playstation Now speed, but by the time it launches more people will have the option of paying for a better connection to support the service.

Region Median Download Speed Median Upload Speed
London 14.3 Mbps 1.4 Mbps
Northern Ireland 13.9 Mbps 1.6 Mbps
North East England 13.3 Mbps 1.1 Mbps
South East England 12.5 Mbps 1.0 Mbps
North West England 12.2 Mbps 1.0 Mbps
West Midlands 11.9 Mbps 0.98 Mbps
East Midlands 11.5 Mbps 0.97 Mbps
Yorkshire and Humber 10.6 Mbps 0.92 Mbps
South West England 8.9 Mbps 0.86 Mbps
East of England 8.8 Mbps 0.86 Mbps
Scotland 8.1 Mbps 0.82 Mbps
Wales 6.6 Mbps 0.76 Mbps
UK Regional Broadband Speeds via thinkbroadband speed test, analysed January 2014


Posted by michaels_perry over 3 years ago
It appears that these speeds are heavily biased by those connections using fibre. Those of us who don't have that option and have to use ADSL, as fibre has not been installed and in some cased is unlikely to be available for some years yet, are left with much slower connections. Some of these are deliberately, in my opinion, being held slow artificially by ridiculously high noise margin settings in exchange equipment.
Only by separating out the fibre and ADSL results will you be able to show a truer picture.
Posted by andrew (Favicon staff member) over 3 years ago
These are regional figures so will of course include all connections across a region.

Remember also we are displaying median speeds, rather than mean which is the statistic that is easily skewed by a small minority getting a fast speed.
Posted by csimon over 3 years ago
Ofcome have recently announced that "the UK average peak-time download speed is 14.2Mbps", which tends to suggest they've been looking at London only. See
Posted by csimon over 3 years ago
Ah sorry, this is the same press release that was linked above, just from a different site!
Posted by chilting over 3 years ago
The speed test figures look pretty accurate to me. Clearly they are pulled down by areas that BT still have to upgrade. Such as the West Chiltington exchange in Sussex. The only surprise is that the South West is so low in the rankings with all the money spent in Cornwall.
Posted by gah789 over 3 years ago
I think that one should take these figures with a pretty high degree of scepticism. True, the median is more representative than the mean, but that is only the beginning of your statistical problems. The main difficulty is handling repeated speed tests by people worried by (relatively) slight variations in speed. This introduces a large component of what statisticians call "selection bias". I know I have a slow broadband service. There is no point in repeatedly testing it.
Posted by andrew (Favicon staff member) over 3 years ago
The joy of the difference between median and mean figures.

e.g. our mean for London is 24.9 Mbps, but median is 14.3 Mbps.
Posted by andrew (Favicon staff member) over 3 years ago
@gah789 we take into account repeated tests, e.g. that obsessive who tests they 100 Mbps link every 8 hours.

If people think the stats are wrong, then the answer is do a speed test and make sure your postcode is represented.
Posted by jumpmum over 3 years ago
It would be interesting to see if the different parts of Wales show a significant difference now so much of Gwynedd and Anglesey have been enabled. Itmay be you are light on speed tests in some parts. ( looking at your map).
Posted by andrew (Favicon staff member) over 3 years ago
The areas light on speed tests are also often the areas with less people living in them.
Posted by PhilCoates over 3 years ago
@andrew - presumably because their connections are poor, they know that and don't want to keep being reminded by testing!
Posted by andrew (Favicon staff member) over 3 years ago
@philcoates I believe we have a better representation of the rural picture than the Ofcom data which appears to extrapolate 2009 data at times.
Posted by csimon over 3 years ago
@jumpmum: "It would be interesting to see if the different parts of Wales show a significant difference now so much of Gwynedd and Anglesey have been enabled." There is of course an increase in speeds across Gwynedd but coverage still patchy. On TB's broadband map, a straight-ish line of about 10 miles from Caernarfon reveals tests of 17.48M, 1.36M, 2.5M, 40.07M, 10.29M, 0.75M, 0.39M. Spot where the exchanges are!
Posted by gerarda over 3 years ago
@andrew It is a pity Ofcom do not use independent data like yours as setting a room full of monkeys in front of PCs with spreadsheets would produce a better representation of rural speeds than Ofcom does
Posted by Gadget over 3 years ago
Doesn't Ofcom use the independent samknows boxes to test speeds?
Posted by andrew (Favicon staff member) over 3 years ago
They use boxes, but then take the data from them to extrapolate a UK average, by weighting the various results to compensate for line lengths etc and relative popularity of all the providers.
Posted by gerarda over 3 years ago
Samknows reported to Suffolk County Council in 2011 0r thereabouts that ADSL was universally available in the county so there is something wrong with their data analysis.
Ofcom ignore postcodes with limited or no data thus excluding all not and a lot of slow spots
Posted by WWWombat over 3 years ago
Ofcom have two main reports for speeds.

One they run at 6-monthly intervals, using a few thousand SamKnows boxes. These give a very detailed picture of a few lines, including download speeds, but the panel is self-selected, and because of data usage won't be people on packages with restricted packages.

The other is the annual infrastructure report, with data from all postcodes (save those with too few lines), with sync speeds mostly reported by ISPs but not download speeds or other details.

Two reports. Two very different types of results. Two different average speeds.
Posted by WWWombat over 3 years ago
Doh. "Packages with restricted allowances" in case it wasn't clear
Posted by stanmor over 3 years ago
Another possible distortion source is the increase in tablet use. I now use my iPad for everyday, only go to the PC for specific activities. So as I test what I am using and your speed test doesn't work on iPad (needs Flash), my figures wont be in your survey. I guess that applies to many people and many datasets
You must be logged in to post comments. Click here to login.