Snails, Gazelles, Damn Lies and Average Broadband Speeds
Telco 2.0's Senior Analyst Keith McMahon is not usually an angry man, but today he's uncharacteristically incandescent with rage. The reason is that he has analysed the latest stats from the UK's regulator OFCOM and he smells a rat. Or more accurately, a mixture of snails, tortoises, humans, gazelles, and twisted statistics. Here, only lightly edited to protect more sensitive readers, is what he shared with the Telco 2.0 team this morning.
Another day brings another bunch of misleading statistics released by OFCOM. The headline today was "UK average broadband speeds up to 12Mbit/sec". This headline figure, which was lazily lapped up by the majority of the British press as a sign that things are hunky-dory, hides a lot of truth about the underlying state of the British broadband market.
The chart above and vast qualification adequately describes the real state of the UK broadband superhighway. Basically we have three lanes: tortoises (between 2Mbit/sec and 10Mbit/sec), humans (between 10Mbit/sec and 30Mbit/sec), and gazelles (above 30Mbit/sec). The tortoises and humans haven't improved at all - the gazelles have merely pulled away. Using the speeds above, one gazelle carries the same weight as five humans and ten tortoises in calculating the averages.
Virgin Media announced in March 2012 that it was effectively tripling its speeds at no extra cost. The tiers are now 30Mbit/sec, 60Mbit/sec, and 100Mbit/sec.
The impact on its base has been transformational - humans turned into gazelles, at no extra cost, without the need for a truck roll. The chart above shows the situation at Q3. By Q4, Virgin Media now had 2.2m gazelles on their network, which is around 50% of their base. A truck roll and more money are required for a customer to transition to the Openreach gazelle network, which almost certainly means adoption is much slower, but BT conveniently do not release consumer adoption figures.
OFCOM's estimate of the split between gazelles, humans, and tortoises are shown above. But where are the snails (below 2Mbit/sec), which count for 1% of the market? They are conveniently not monitored anymore. I'm not sure of the logic for this, but it certainly helps to keep the charts looking good.
There are some other tricks which have been included not only to improve the presentation but also keep the averages up. The first trick is ignoring rural customers for the tortoises' comparison. Market 1 is defined as where BT has no unbundling in the exchanges. When OFCOM announced the market definitions, this accounted for 11.1% of UK homes. As these tend to be in rural areas with long copper lengths and therefore lower DSL speeds, the snails appear less sluggish then they really are.
The second trick is ignoring copper loops of over 5km for the humans' comparison. This is another 15% of UK homes, although there will be a substantial overlap with the Market 1 homes. Again, this only ups the human average.
The third trick is to only include on-net customers for Sky, TalkTalk, O2 and Everything Everywhere, i.e. where they have been unbundled. Theoretically, this could be either a positive or negative to the calculated averages, but the mere fact that they have been excluded from the sample raises my eyebrows. In fact, I'm not sure whether this has really happened, as Everything Everywhere figures are included in the detail of the report. As far as I was aware, Everything Everywhere gave up its LLU a couple of years ago and ported the base onto BT wholesale unbundled products.
The fourth trick is that the gazelles are overrepresented in the samples, which may be self-selecting. This, of course, ups the average. This self-selection was my main issue when OFCOM announced the broadband speed project. Basically, the people who would volunteer to have an extra piece of CPE in the home monitoring their connection are tech-savvy users. The type of users who insert the correct filters to the main phone line and try to achieve the best possible speed for their connection.
This is not representative of the average UK broadband user, who probably connects the router to the most convenient phone extension in the home, and thus has reduced speed. The fact that more of the sample are upgrading to superfast broadband only reinforces my belief that the sample is not representative of UK broadband users.
I know the founder of Samknows, the company that OFCOM subcontracts the project to. The original project was conceived from the belief that the UK consumer deserves to understand the true performance of the UK broadband providers, rather than accept the advertised headline speeds. The project also sought to educate them that peak hour speeds, latency, and packet loss were just as important as absolute speeds.
Today, I feel that the project has been hijacked and exploited by OFCOM for political propaganda - "look at how fast our average broadband speeds are". The truth is that our broadband highway has four lanes for snails, tortoises, humans, and gazelles, and speeds are not significantly changing within these lanes. What has happened is that the company with the best overall network, Virgin Media, has decided for competitive reasons to upgrade speeds to a large part of its customer base. Good for Virgin Media and good for their customers. Meanwhile, the rest of the UK awaits BT Openreach to upgrade its crumbling copper network. The politicians have a great headline grabbing message, but most of the UK remains unaffected.
Brave readers can find the full report here: http://stakeholders.ofcom.org.uk/market-data-research/other/telecoms-research/broadband-speeds/broadband-speeds-nov2012/
For what it is worth, I am human, on an ADSL2+ LLU network achieving speeds of around 3.5Mbits/sec. I'm gutted.