Skip Navigation

Broadband speed test results for December 2015
Monday 04 January 2016 11:00:57 by Andrew Ferguson

Another year has rolled by and as January starts and everyone struggles back into the standard weekly routine we have the results from our monthly speed test analysis. For those less interested in individual providers the results for Q4/2015 across the various parts of the UK have already been added to our speed and availability checker which also includes maps of postcodes below 2 Mbps and 10 Mbps.

Large Provider Fibre Based Connection Speed Tests December 2015
Provider Median Download Mean Download Median Upload Mean Upload
FTTC Overall (excludes Virgin Media) 28.4 Mbps 29.4 Mbps 6.6 Mbps 7.3 Mbps
BT 30.7 Mbps 31.7 Mbps 8.1 Mbps 8.5 Mbps
EE 28.1 Mbps 27 Mbps 8.1 Mbps 7.3 Mbps
Plusnet 30.6 Mbps 31.5 Mbps 6.9 Mbps 8.4 Mbps
Sky 24.8 Mbps 24.6 Mbps 6.7 Mbps 6.5 Mbps
TalkTalk 28.1 Mbps 28.8 Mbps 1.8 Mbps 4.3 Mbps
Virgin Media 41.3 Mbps 49 Mbps 4.8 Mbps 5.7 Mbps
Vodafone 29.1 Mbps 31.1 Mbps 8.9 Mbps 9.5 Mbps

Vodafone has joined the group of providers with its FTTC service and the usual small fluctuations in the speed of the FTTC services and no major movement in the average speed for Virgin Media where upgrades are underway. The split of FTTC users we see is that around 20% are on the up to 76 Mbps variant (up to 19 Mbps upload), 70% are on the up to 38 Mbps version (up to 9.5 Mbps upload) and the cheapest version sold by Plusnet and TalkTalk with an up to 2 Mbps upload accounts for 10% of the speed tests seen on FTTC based connections.

ADSL/ADSL2+ Connection Speed Tests December 2015
Provider/Area Median Download Mean Download Median Upload Mean Upload
All Providers 5 Mbps 6.3 Mbps 0.6 Mbps 0.6 Mbps
BT 4.4 Mbps 5.9 Mbps 0.4 Mbps 0.5 Mbps
EE 4.2 Mbps 5.8 Mbps 0.6 Mbps 0.6 Mbps
Plusnet 4.9 Mbps 6.1 Mbps 0.5 Mbps 0.5 Mbps
Sky 4.9 Mbps 6.4 Mbps 0.6 Mbps 0.6 Mbps
TalkTalk 5.3 Mbps 6.4 Mbps 0.6 Mbps 0.6 Mbps
Rural ADSL 3.4 Mbps 4.3 Mbps 0.3 Mbps 0.4 Mbps

The ADSL/ADSL2+ speeds are almost a copy and paste of the previous month with just minimal changes. The overall provider speed table relies on the popularity of the competing technologies, and the split of tests in December 2015 was such at 44% of tests were identified as on an ADSL/ADSL2+ connection, 24% on FTTC and 20% on cable broadband, which is roughly in line with reported take-up figures in the last financial quarter.

The 20 Fastest UK Broadband Providers in December 2015
(ordered by median speed)
Smaller providers without enough geographic data samples are not included
Provider Median Download Mean Download Median Upload Mean Upload Download Speed of top 10%
Hyperoptic 54.7 Mbps 116.5 Mbps 36.7 Mbps 77.1 Mbps 343 Mbps
Virgin Media 41.3 Mbps 49 Mbps 4.8 Mbps 5.8 Mbps 101.8 Mbps
Gigaclear 40 Mbps 57.6 Mbps 32.4 Mbps 78.6 Mbps 102.4 Mbps
Vodafone Broadband 19.4 Mbps 22.2 Mbps 6.2 Mbps 7.3 Mbps 46.1 Mbps
EE Mobile 13.1 Mbps 18.6 Mbps 1.4 Mbps 4.2 Mbps 41.2 Mbps
IDNet 12.9 Mbps 20.7 Mbps 0.8 Mbps 5.3 Mbps 53.1 Mbps
BT 12.9 Mbps 18.6 Mbps 1.4 Mbps 4.8 Mbps 39.1 Mbps
Zen Internet 12.8 Mbps 22.5 Mbps 1.1 Mbps 6.3 Mbps 61 Mbps
XILO Communications 12.5 Mbps 20.5 Mbps 0.8 Mbps 5.6 Mbps 53.4 Mbps
Claranet SOHO 12.3 Mbps 23.3 Mbps 0.9 Mbps 10.2 Mbps 69.7 Mbps
AAISP 10.2 Mbps 19.5 Mbps 1.9 Mbps 6 Mbps 56.7 Mbps
Vodafone Mobile 9.7 Mbps 13.5 Mbps 1.3 Mbps 3.1 Mbps 31.4 Mbps
Relish 9.4 Mbps 15.5 Mbps 1.5 Mbps 2 Mbps 35.7 Mbps
Europasat 8.3 Mbps 8.4 Mbps 0.1 Mbps 0.2 Mbps 14.9 Mbps
Plusnet 7.9 Mbps 15.3 Mbps 0.8 Mbps 3.9 Mbps 37.6 Mbps
02 Mobile 7.8 Mbps 12 Mbps 2.8 Mbps 8.4 Mbps 30.5 Mbps
KC 7.3 Mbps 21.1 Mbps 2.6 Mbps 4.9 Mbps 65.1 Mbps
Sky 7.1 Mbps 11.3 Mbps 0.9 Mbps 6 Mbps 28.5 Mbps
Daisy Wholesale 7 Mbps 15.3 Mbps 0.7 Mbps 3.5 Mbps 38.1 Mbps
TalkTalk 6.9 Mbps 12.4 Mbps 0.9 Mbps 8.8 Mbps 33.3 Mbps

As usual Hyperoptic is atop the table, but Gigaclear beaten to second place by Virgin Media may surprise some, but once you realise that the most popular products at Gigaclear are the 50, 100 and 200 Mbps products rather than Gigabit you can see why its almost a head to head fight with Virgin Media who also sell 50 Mbps, 100 Mbps and 200 Mbps products.

4G from EE is making its mark by climbing the table in December and is showing that mobile can provide decent speeds when combined with the right handset. Sky and TalkTalk may be down at the foot of the table, but this is really down to the way the various products are promoted, for BT we see an almost equal amount of ADSL and FTTC customers, but for Sky and TalkTalk there was almost three times as many ADSL users as fibre based users. This is not just a skewing from our visitor numbers but is also borne out by the financial reports from the providers and with the plethora of cheap ADSL2+ deals will continue for a long time.

Of the providers who did not make the top 20 the ones missing due to their small sample size were B4RN (48.8 Mbps median download), IFNL (38 Mbps), Metronet UK (36 Mbps), Wight Fibre (23.5 Mbps), Wessex Internet (20.8 Mbps) and Kijoma (20.7 Mbps).


Posted by mantequilla 10 months ago
You mention sample size regarding B4RN and others, what is the sample size for the larger players and what sort of a percentage is this of the UK population?
Posted by andrew (Favicon staff member) 10 months ago
If you want to gain an idea of the sample sizes involved look at with the speed test layer turned on.
Posted by TheEulerID 10 months ago
The statistics of sampling are such that smaller population groups need a higher percentage of measurements than for larger groups to get the same degree of accuracy.
However, that's all based on random sampling. For self-selecting sample sets such as TBB has to use, there is always a vulnerability to skewed results.
Posted by TheEulerID 10 months ago
However, as the month-on-month figures do not seem to be subject to much erratic movement (at least for the large players), it is almost certainly a good indication of overall trends in speeds.

nb. I would expect those ISPs with very small population groups to be much more erratic in results.
Posted by chilting 10 months ago
I have mentioned before just how bad the upload results are for Satellite Broadband. This month Tooway Direct is still stuck on 0.1 and 0.2Mbps
However, the download has fallen from around 13.3 to 8.3Mbps and the speed of the top 10% from 19 to 14.9Mbps compared to November 2015.
I guess that it is to early to attribute the fall to the BDUK USC offer but if the trend continues...
Posted by Blackmamba 10 months ago
Hi Broadband Watchers.
I feel the most important results are the new 2 and 10 map results but I have found that many are not correct but it looks like they will be removed over time as there seams to be a lag time. Eg some post code results are very close to the new Cabs and the Exchanges.
Posted by andrew (Favicon staff member) 10 months ago
If no one reports errors then errors tend to remain.

The amount of changes means that if people want roll-out updates to be done daily then we would have to charge for each lookup so that people dedicated to updated the maps can be employed.
Posted by cyberdoyle 10 months ago
I think your speedtester is a bit wonky. I tried it twice today and it said I had less than 10 meg upload whereas in fact I have over 900 mbps upload.
Posted by cyberdoyle 10 months ago
tried your speedtester again this morning in case it was overloaded last night, but it still only testing at 400mbps down and 200 or so upload. Just did another test to dsl reports tester (pure html as opposed to flash) and that gives me over 900mbps symmetrical. - maybe you should ask BT for a better server? or am I doing something wrong?
Posted by andrew (Favicon staff member) 10 months ago
Nothing wrong with the tester, the fact other people with Gigabit services hit reasonable speeds shows nothing wrong.

As for the flash comment, we run two testers, one is flash and one is not.

dslreports uses around 32 servers to deliver your 'good report' from various countries, which means even if an ISP is having a bad day it is able to report a good speed.
Posted by andrew (Favicon staff member) 10 months ago
Example other test:

One of the latest B4RN tests:

The variable speeds suggest maybe a route between B4RN land and London is not so good. So maybe look at peering link sizes in Manchester.
Posted by Dixinormous 10 months ago
'maybe you should ask BT for a better server?'

Posted by mantequilla 10 months ago
'to gain an idea of the sample sizes'. OK central Oxford around Merton College has 8 results. Really? Let me try again, what number of results do you have as a percentage of homes in the UK?
Posted by mantequilla 10 months ago
Trying to count the number of results on the map is hopeless. Also I don't know how many premises there are in a given map so your suggestion doesn't really give me the sample size. Also results are disabled at a certain zoom level anyway, I haven't figured out when as yet. Just tell me how many samples you have.
Posted by Blackmamba 10 months ago
Hi Manteguilla.
If you look on Openreach and type in post code on the fibre access section you will get the addresses tied to that area post code it will also give the customers that are unable to get service due to ( long line or other reasons ). Remember each post code has different number of homes (addresses).
Posted by andrew (Favicon staff member) 10 months ago
The number of results is commercially confident and is gathered from tests run on a number of sites that use our tester technology.

Posted by mantequilla 10 months ago
Just as an experiment, I took a random city, population 107K, 42K households. Number of thinkbroadband samples 1550 (gleaned from map), gives a sample size of 3%, less due to repeated experiemts on same connection. That is pretty small.
Posted by andrew (Favicon staff member) 10 months ago
And Ofcom uses 2,000 data points for the whole UK and extrapolates from that.

No modelling used in the figures you can see, and if the sample was unrepresentative we should see wild swings each month.

Posted by andrew (Favicon staff member) 10 months ago
Should also point out that not all our speed test partners data appears on the map, but we do have it for data analysis purposes.
Posted by mantequilla 10 months ago
Someone else being worse is hardly relevant. So 3% is about right then or perhaps a little on the high side?

The opinion polls in May were not representative but were remarkably stable and wrong.
You must be logged in to post comments. Click here to login.