Ofcom reports UK average download speed in November 2017 as 46.2 Mbps
The annual Ofcom fixed line broadband performance report has been published with a deep analysis of the performance from the UK wide SamKnows testing panel.
The headline figure is that the UK now has an average download speed of 46.2 Mbps (average upload speed 6.2 Mbps) measured in November 2017 considerably higher than a year ago when it was 36.2 Mbps.
For those not familiar with how the performance report works it relies on up to 5,000 testing boxes sent out to the public which test that line multiple times in the day to build a large profile of performance over time for that line, this is combined with the line characteristics and modelling to arrive at the various conclusions. Though it does seem a bit confusing when some data outputs use a panel base of 4,861 and others a panel base of just 1,369.
The report does highlight that the testing does not include the effects of Wi-Fi which is something our crowd sourced results will, and we don't do any modelling on our average speeds or the different technology speeds we report each month or feature on our statistics site. In Q1 2018 for the UK we recorded an average download of 30.4 Mbps (6.4 Mbps upload) compared to 25.8 Mbps down (5.1 Mbps up) so we agree with Ofcom that there is significant improvements in the speeds seen but factors such as Wi-Fi and the tendency for online speed testing to attract those who may be having problems largely explains why there is the difference in the two datasets e.g. if we eliminate a chunk of the obviously broken broadband connections testing the Q1 average rises to 31.8 Mbps and other filtering and modelling would get this even higher.
For those in rural areas there is the unsurprising conclusion that with lower superfast service availability, combined with longer lines for ADSL/ADSL2+ services that rural areas see lower speeds but this is not always the case e.g. West Berkshire when we split it into its urban and rural components has a very similar median download speed of 26.5 Mbps and 24.5 Mbps but then rural West Berkshire is only 0.1% below the level of superfast coverage of urban West Berkshire. Rural West Berkshire has a lot of Gigaclear coverage, which is reflected in the mean upload speed of 21.8 Mbps versus an urban mean upload of 6.9 Mbps.
The performance during the 8pm to 10pm window is of most interest given the changes that are now just two weeks away and it is interesting to note that our figures seem to be more inline with the minimum reported by Ofcom, this is probably down to adding to our analysis the median speed of each individual test, rather than a mean or top quartile (some analysis presented for the UK by others than Ofcom seems to use a top quartile). Ignoring the differences between our crowd based analysis and the Ofcom system the ASA and CAP may well need to pro-actively ask those providers who claim median speeds above the Ofcom figures to present the data for full analysis, especially if we end up in the scenario where all the old up to 38 Mbps services claim a figure above the 33.5 Mbps.
Ofcom has noted there was an improvement for Virgin Media customers during the 2016 to 2017 period and we noted this particularly during the summer of 2017 after their appearance on BBC Watchdog, though while the Ofcom report mentions the roll-out of the 300 Mbps service and capacity upgrades they do omit the changes in the headline room from 10% to 15% on the cable products which looks to be in preparation for the ASA average speeds rule. The move by BT offering free speed upgrades to Infinity 1 customers onto a 76 Mbps tier was handled the same way by Ofcom as we have i.e. those whose service has been upgraded count as up to 76 Mbps customers.
The single stream performance is something that we share in common with Ofcom and we don't know what speed they consider as needed for supporting UHD but they suggest 18% of ADSL2+ customers can manage to stream UHD which seems suprising since we have found that 4K/UHD content tends to only reliably run once connection speeds are above 24 Mbps. What the single stream results may surprise some, the up to 100 Mbps cable service is out performed by up to 52 Mbps, which supports our own analysis that single thread performance can be less than ideal, since what the Ofcom report is saying is a fixed connected speed service of 100 Mbps can perform worse than an ADSL2+ service if you are on a congested area.
This difference in performance where faster download speeds does not always equate to better performance is reflected by BT Infinity 2 (up to 76 Mbps) scoring a latency of around 12 ms versus 18 to 20ms for 100 Mbps Virgin Media service. The odd quality score that we consistently give EE seems to be reflected by the Ofcom report too with EE recording a time of 400ms versus around 180 to 250 ms for other basic superfast packages.
For the Ofcom staff reading in the glossary of the 91 page report they may want to correct how they describe the single thread test i.e. 'A test involving the download of a single file: Single-thread tests typically record faster speeds than multi-thread tests, in particular for higher-speed connections' since one actually expects mult-thread testing to be faster.