Telstra has always dominated the telco market. Today, rather than relying on its incumbency, the telco is holding onto its lead through the strength of its products. The latest data from ZDNet Australia's Broadband Speedtest not only shows how Optus is vying with Telstra for leadership in the broadband speed stakes, but also how the two of them are way ahead of the pack.
Every day more than 5000 people across the globe take the ZDNet Broadband Speedtest. Users get to test the speed of their connections and we get a wealth of data about the speed and performance across a variety of networks. Over the six months from 1 May to 31 October 2011, almost a million tests were run in Australia — that's not a bad sample size to be working with.
Connection speeds of just under 100Mbps were recorded, but overall such high speeds were rare. Only 0.3 per cent of tests (3400 of them) recorded a speed exceeding 80Mbps. Even so, the vast majority now enjoy speeds above 1.5Mbps and close to a third are faster than 8Mbps. In this report we'll show how quickly adoption of higher speeds is increasing, and how Telstra and Optus are leading this charge.
Even limiting our analysis to data collected over the past six months, we can see how connection speeds are increasing. Those experiencing speeds in excess of 20Mbps increased steadily from 4.9 per cent of all tests in May to 5.9 per cent in October. The increase was less marked, but still evident, in the 8 to 20Mbps range. The bulk of users remain in the 1.5 to 8Mbps range, although this has slipped slightly from 55 per cent in May to 53.5 per cent in October. Still, that's a long way from the days when life maxed out at 1.5Mbps. Now only 17 per cent fall into this "slowcoach" bracket.
Another interesting bit of information our test collects is the service provider being used by the tester. We all know that Telstra has a clear lead when it comes to broadband subscribers. What's surprising is that in the six months from May to October, TPG beat OptusNet to place second (fourth-place iiNet scored roughly half the tests of TPG). Providers such as Netspace, People Telecom and Unwired accounted for a fraction of 1 per cent of total tests.
Now it's dangerous to assume that these figures are truly representative of market share. For a start, 19 per cent of tests could not be allocated to a specific provider (due to changes in IP addresses). We should also be aware that some service providers attract a more "techie" crowd — a crowd more predisposed to periodic speed testing of their connections. There could also be a bias if a provider were to offer a lower standard of service, drawing more people to perform periodic speed checks.
That said, tracking such a large amount of data should be able to provide an indication of shifts in the share of each provider. BigPond's dominance, for example, has slipped slightly — falling from 36.1 per cent of tests in May to 35.3 per cent in October. Optus suffered a 2 per cent drop, from 13.5 per cent to 11.5 per cent. TPG and Internode also fell around 1 per cent, while iiNet and Dodo picked up a similar amount of share. We can't be entirely sure that this is indicative of the market, but it seems to hint that the big players aren't necessarily just getting bigger.
So here's the crux of the Speedtest: what speeds are users actually getting? There are, of course, several caveats. Results could depend on what plans people are on and the quality of their connection. For example, if one provider averages a higher speed than another, it could be because the first provider doesn't offer lower speeds.
The upshot is: we can't equate fast speed results with a better-quality network. That said, if we look at the results recorded, there's no denying that Optus and Telstra are way out in the lead. They are offering average speeds of 9.6Mbps and 8.8Mbps, respectively. That's a fair way ahead of Internode, TPG and iiNet who are struggling to average 6Mbps.
While Optus might be the winner in the average speed stakes, let's remember that Telstra has many more customers. As the graph shows, we're not just seeing variations in speed — performance (or Time To First Byte) results also differ enormously. NetSpeed offers the lowest TTFB at an average of 492 milliseconds (based on 671 speed tests) whereas Unwired lags (so to speak) well behind the rest at 1872 milliseconds (based on 638 tests). The good news is: if you are with one of the larger providers you will enjoy comparatively low TTFB — Optus, BigPond, Internode, iiNet and TPG all fall below an average of 700 milliseconds. Primus is slightly above that level (721 milliseconds) and Dodo lets the side down with 1119 milliseconds (based on 18,457 tests).
* Initially, in the performance charts on this and the following page, we used the term "latency". As this score is not the same as technical latency, we have corrected the references to read "Time To First Byte" (or TTFB) — a measurement that represents the duration from the user making an HTTP request to the first byte of the page being received by the browser. We apologise for the confusion — the measurements still indicate the relative real-world responsiveness of the various providers.
Just over 8 per cent of the 980,000 Australian speed tests were performed from work. As you might expect, it was harder to detect the provider in many of these cases — in fact, more than half of all tests came back "unable to detect". Of those that were identifiable, Connexus Internet came out on top, averaging 21.6Mbps for each of the 288 tests run. For the results where the provider was not identified ("other" networks — which accounted for 43,122 tests) the average was 11.1Mbps, still quite a way ahead of Telstra BigPond at 9.4Mbps. Of course, many of those other networks could well be provided by Telstra, or Optus, or any other provider, but if the IP range is addressed to the corporation concerned we have no way of knowing.
Let's also remember that the Connexus results are averaged over a small number of users, whereas the Telstra BigPond and OptusNet figures come from 14,297 and 4089 tests, respectively.
The growth of wireless
The number of speed tests performed from mobile devices has also increased over the last six months. While mobile access accounted for 2 per cent of speed tests in May, it accounted for 2.9 per cent in October. In the last six months mobile users averaged a connection speed of 2.74Mbps.
An analysis of the service providers used for each speed test would seem to indicate that the major players (Telstra BigPond and OptusNet) are rapidly increasing their share of the mobile broadband market. TPG is also seeing some growth (albeit from a lower base), while other major ISPs are struggling to hold their share. These figures could be influenced by the vagaries of IP addressing, of course; for example, resellers of one network could be included in the carrier's figures. That said, it wouldn't be a surprise to anyone if network operators did what they could to promote their own retail offering at the expense of their wholesale customers.
You'll notice Vodafone is conspicuous by its absence — it is no doubt included in the massive "other" figure for wireless, which accounts for 30 per cent of the market.
So why are the big carriers doing so well, and the others not so? Well, for Telstra it's almost certainly a question of speed. Over the six months from May to October 2011, the average speed for a Telstra wireless connection was 4Mbps. Although iiNet and Internode aren't far behind, it was Optus, which averaged just 2.1Mbps, that seemed to be gathering the most share. That would seem to indicate that, beyond a certain speed, it's price that influences purchase choice. (Unless, of course, Optus customers are so unhappy with the speeds they're getting that they keep going back to the Speedtest to check it!)
* Initially, in the performance charts on this and the previous page, we used the term "latency". As this score is not the same as technical latency, we have corrected the references to read "Time To First Byte" (or TTFB) — a measurement that represents the duration from the user making an HTTP request to the first byte of the page being received by the browser. We apologise for the confusion — the measurements still indicate the relative real-world responsiveness of the various providers.
What's clear from this wealth of data is how quickly the dynamics are changing — speeds are increasing, market share is shifting and mobility is rising, month by month. This has to be encouraging for the user, who should be experiencing a noticeable improvement in service. Of course, the data also suggests there is a huge variety in the quality of service, with some providers showing lower speeds and higher TTFB than others. It demonstrates the need to maintain quality if providers want to stand a chance in such a cut-throat market.
Price reductions will bring short-term gains, but is such a tactic sustainable? Customers will continue to expect quality improvements. Dodo, for example, increased its share while increasing speed from an average of 3.7Mbps to 4.4Mbps, and while reducing TTFB — clearly an issue — from an average 1343 milliseconds down to 918 milliseconds. Compare that to Optus. Its TTFB and speeds are better, but Optus' share of speed tests fell 2 per cent over the six months of the analysis, during which time TTFB increased from an average of 625 milliseconds in May to 690 milliseconds in October (all over fixed connections).
These are still a lot better than the Dodo figures, but Dodo has always sold on price. Today, these figures seem to show, price is important, but it needs to be matched with faster speeds and lower TTFB if you are a service provider that wants to be in with a fighting chance.