We published the basic table of the speeds from the various UK broadband providers where we have enough data to produce some meaningful statistics on our main site and now are using this blog to give people a deeper insight into the results.
Latency is something we record from our testers and the above bar chart shows the median latency for the twenty best performing providers we have data for. An important point to make is that this latency measure does not totally equate to ICMP ping times, since it is measured using the round trip time for a small TCP packet, but in terms of measuring relative latency performance it will show the better performing providers. AOL and Pipex will vanish from this chart when the December 2014 results are published as their IP blocks are now merged into the TalkTalk group. O2 Legacy will also be absorbed into the Sky results.
Virgin Media Results
Virgin Media appears to have had a peak speed for 2014 during the summer months, but the median speed appear to have stabilised in November 2014. In theory if Virgin Media can reduce the level of complaints about peak time speeds dropping significantly we should see wide speed plateaus at the various connection speeds.
For those wishing for more upload speed at Virgin Media it is worth comparing the upload and download profiles, as it shows that while a reasonable number are on the fastest 152 Mbps package only a small proportion are getting an upload speed that is 10% of the download. The 12 Mbps upload speed looks a lot more common.
The advantage to publishing these profile charts is that it allows people to visualize the median speeds, quartile and percentile speeds.
Zen Internet Results
The popularity of the Zen FTTC services is clear from the profile with around 43% of the tests coming from an FTTC based connection.
The sharp uplift in speeds that FTTC offer is very obvious with Zen and there appears to be no preference between the 40 or 80 Mbps FTTC based service.
BT Consumer / Retail
With the BT retail service you can see a distinct levelling at the maximum test speed of the up to 40 Mbps FTTC wholesale product, it would appear that the pricing differential of Infinity 1 and Infinity 2 mean that we are not seeing a smoother drop off in speeds over distance compared to Zen Internet.
While it may irk some broadband campaigners the reality of life is such that households are generally happy with the broadband if they stream video, HD preferred but the desire and need for 4K streaming is still some way off. How far away depends on the wider UK economy and how quickly the nation will swap out existing perfectly good HD televisions for new UHD models.
We have annotated the PlusNet profile charts to illustrate some interesting little facts. Mainly that the slower FTTC users almost smoothly merge into the realm of the faster ADSL2+ users, and if one is trying to estimate how popular fibre based services are that they should use the cross over point in the upload profile to guide them.
The practice where the PlusNet provisioning portal often caps ADSL2+ users speeds at 448 Kbps is visible, though as some other BT Wholesale based providers show this it is possible we are just seeing the fact that BT Wholesale ADSL2+ has still not rolled out fully across the UK.
KC has often looked different when looking at the speed profile for users, and we are confident that this is down to the high proportion of FTTH in their fibre roll-out which means those who do order faster speeds are more likely to hit the purchased speed (i.e. there is no distance versus speed drop).
Given with all the other providers the proportion of users testing when on faster speeds is relatively high compared to the actual proportion of users declared as buying the faster products, it is a bit disappointing to see a relatively low number of fast results, even if the actual tests are good. One thing holding people back in the KC/Hull area may be that the fibre based products all carry usage limits, and on a 100 Mbps connection a speed test might use up 200 MB of your allowance we may see less people testing.
At first glance the Sky upload profile looks wrong, but if you look at the BT Consumer upload profile there are hints that the upload speeds for the 40/10 service degrade in this bell like manner. Obviously the 2 Mbps upload cap on the TalkTalk service means this is not visible.
The fact that people are trading speed for money in their wallet is again made clear by the lack of people buying the Fibre Pro service at £30 per month. This is not because FTTC does not perform, the charts for PlusNet and BT show that the public seem to not want to pay much of a premium for speeds above what they feel is currently enough.
2015 will be very interesting as should hopefully see for both Sky and TalkTalk evidence of the FTTP roll-out in York and as it will be FTTH based it will be obvious on the speed profiles, unless the pricing is set so that people only buy slower speeds that are swamped. Of course we can do specific analysis on the roll-out areas in York which should help in identifying how popular the service really will be.
EE is showing the classic case of its entry level fibre based service being popular, but the more expensive 80/20 based service is only as popular it seems as TalkTalk and Sky. This is interesting as the new EE TV product means EE is trying to compete with Sky and TalkTalk (and even BT) in the TV/broadband and phone area. Of course EE does have the advantage of being a major 4G operator and we will take a similar look at the speeds of the UK mobile providers in a few months.
TalkTalk is almost on its own in selling the 40/2 FTTC service, the advantage it gives is that it is slightly cheaper at the wholesale level, and running a network where heavy uploaders are probably discouraged from joining due to a slow upload might make it cheaper to provide the service. The good news seems to be if you are on a budget and after faster downloads that the FTTC service from TalkTalk still seems to deliver the download speeds.
The reason why TalkTalk was at the bottom of the table appears to be because the old AOL IP blocks are being used by some FTTC customers, which has reduced the median and mean speeds slightly. We would estimate the difference would move TalkTalk to be on par with Sky and once December is over we will know for sure, as the provider tracking was updated on the 1st December 2014.
If there is a lesson to take from this speed test round-up it is that FTTC based performance is governed as one would expect really on the Openreach FTTC hardware. Also it is clear that speed on its own does not sell, i.e. build it and they will come may be a dangerous assumption, certainly the evidence we see suggests yes people will come but they will try and pay the minimum needed for their needs.
It is likely that Virgin Media will stay at the top of the monthly speed test round-ups for a long time, at least until alternative operators like Gigaclear and Hyperoptic reach a large enough volume of locations tested to make an appearance. In theory the 160,000 or so premises able to order a GEA-FTTP service as their only Openreach superfast broadband option should show up in these results, but we have grown used to people opting for the 40 or 80 Mbps products, rather than pay the £15 to £25 premium that the 200 Mbps and 330 Mbps options require.
We have looked at asking people what type of technology their connection is based on, but the error rate for users answering this question is generally higher than for an automated detection system and by the time the December 2014 results are published we intend to be able to publish ADSL2+ and FTTC speeds separately for the major providers.