Back in August 2013 we published an extensive round-up of speeds from our speed test where people provide a location in the form of their postcode. While that analysis showed the median and mean speeds, there was some who wanted further detail, particularly what the percentile distribution for speeds were, and we have re-run our analysis on the same dataset (the same data was used so people can easily compare it).
The 12 UK regions are plotted and what the chart tells us is that 10% of the speed tests we see from a Welsh broadband user are under 1 Mbps, whereas in London 10% get under 2.5 Mbps. The story at the faster end is that 20% of tests were faster than 20 Mbps in Wales, compared to a much higher figure 37 Mbps for London. In theory if the Universal Service Commitment is implemented properly we should see these worst performers increase, though due to issues like wireless, home wiring, congestion there may still be a few seeing slow speeds. Of course this plot is not a perfect representation of what is available, as there may faster products available to people but for financial or other reasons they do not buy them – and that is a key part of the superfast roll-outs, to drive take-up both in the commercial and Government assisted areas.
It is interesting to see Northern Ireland featuring very well and this down to the level of take-up (thought to be over 20%) and availability for FTTC services, which while not totally universal are possibly closer to what we might expect at the conclusion of the BDUK projects. Before everyone uses this data to say that the £530m of BDUK money is only going to deliver an average speed of 14 Mbps, remember that this is real tests from real connections and not a projection based on cabinet to premises distance, so those electing for the cheapest £2.50 per month services will be pulling the figures down.
This second chart very neatly illustrates the digital divide in the UK, and is the same data but plotted with 1000 Mbps as the maximum speed. We do have people running speed tests at even higher speeds, but all too often people do not provide a postcode. As with our other speed test analysis we do aggregate results for an individual postcode, so that a user who is obsessive about testing their broadband speed does not unduly skew the results with 1000 results at 0.4 Mbps, or similarly with the very ultrafast connections.
Broadband is not just about download speeds, more of us are uploading material, either creating content, or just the more mundane backup of our devices to the myriad of cloud services, so we have done the same plot for upload speeds.
The curve of the distribution is slightly different for upload, and this is mainly because of the big jump in upload speeds that the Openreach FTTC products provide and the small amount of symmetric fibre optic services that are available across the UK.
As a one off dataset these charts are not that useful, but as take-up of FTTC increases and Virgin Media continues its speed upgrade programme we should see the things change and thus we intend to try and return to a regional analysis every 3 to 6 months and fit in some more detailed looks at provider specific speeds in between.
Also by publishing this information independent of Ofcom or any other body we can show the wider picture, as the large national reports can be less accessible, i.e. we have produced these charts in response to requests and are open to suggestions for improvements.