Mobile operators have been told for the last two or three years that they should be fretting about how to reduce network costs when faced with explosive data growth. Everybody in the industry will have seen many examples of the so-called "scissor-graph" showing a curve for traffic and costs going one way and a curve for revenue heading off at 90 degrees.
This picture is flawed for many reasons. We pointed these flaws out two years ago, and our views are no longer controversial.
- Demand is not a force of nature, and open-loop forecasts are, literally, fantastical.
- Growth is constrained by the capital intensity required to deliver it.
- Operators can, if they choose to do so, very effectively pull pricing levers to control traffic.
- New users will dilute any growth in average usage per device.
- There were already signs two years ago that the shifting balance between smartphone and mobile broadband would result in lower growth of mobile data traffic. Growth in Western Europe was 61 per cent in 2011, and every sign indicates it will be far lower in 2012.
- Demand for mobile broadband is cyclical, and a way of quickly monetising excess capacity, but as networks fill up operators will ideally wish to swap mobile broadband traffic out for higher-value, lower-volume handset traffic.
- It is easier for operators to create an artificial spectrum crisis by exaggerating demand than to improve the utilisation of what spectrum they have.
And last, but by no means least:
- Wi-Fi is the default network for most smartphone and tablet traffic. Wi-Fi is not a work-around for cellular in the home and in uncontested public spaces--it is the default option--although it is a poor substitute for outdoor mobile data. The most data-intensive traffic will tend to get consumed indoors, because that's the natural place to consume it, not because it is the only place with good enough connectivity.
All of which is to say that the data wave, or capacity crunch, or whatever one chooses to call it, isn't as big as had been predicted. But the real problem could be that it isn't big enough. Instead of worrying about the impact of a mobile data explosion, operators should worry about making it happen in the first place. Current and future trends as we see them indicate that there is not enough growth in mobile data to stop the mobile industry in most developed countries from contracting. Coupled with some loss of core voice/messaging revenue to over-the-top players, and some erosion of their position in device distribution, this contraction could be severe.
Too little, not too much, data growth
In our recently published Wireless network traffic worldwide: forecasts and analysis 2012–2017, Western Europe has the lowest growth rate in mobile data out of eight global regions. We forecast that mobile data in Western Europe will grow at a CAGR of just 29 per cent from 2012 to 2017, equivalent to a growth multiple of 3.6. At a global level, we predict that mobile data will grow by a multiple of 5.5, equivalent to 41 per cent CAGR, a little ahead of what we predict for Internet traffic as a whole.
Figure 1: Growth multiples for mobile data traffic, by region and worldwide, 2012–2017 [Source: Analysys Mason, 2012]
These growth rates are not close to the "doubling-every-year" forecasts that tend to get heard a lot, but they are nonetheless decent enough growth rates. So what is the problem?
Costs too low, not too high
Over the last decade or so, improvements in spectral efficiency have contributed to a decline in mobile network unit costs (cost per byte) equivalent to about 30 per cent every year, and we expect this trend to continue. This means that if volume growth is higher than 30% per cent (1–30 per cent) – that is, 42.8 per cent or ×5.9 over 5 years, then overall costs rise; if growth is lower, costs fall. Our forecasts indicate that growth will be lower than this in developed regions.
In an uncompetitive market, falling costs would mean rising margins, but in a mature, competitive market such as Western Europe they mean a contracting business. In other words, it's not getting dangerously expensive to cater for demand; it's getting worryingly cheap to transport what little demand there is.
So what can mobile operators do about it?
Mobile data has an imperfect but cheaper substitute, Wi-Fi, which makes demand stimulation difficult. Consumers have got very familiar with using Wi-Fi as an alternative to mobile, and in most respects appear to prefer it. In order to stimulate demand for mobile data, operators could do the following.
- Offer bigger, fatter packages on handsets. Our on-device tracking app showed that smartphone users in the United States tended to use proportionately more cellular (49 per cent as opposed to 38 per cent in Western Europe). However, these packages need to be firmly embedded in a higher-value bundle of device, data and voice; cellular data will almost always be more expensive than Wi-Fi. The trend in both regions, though, is toward a higher share of Wi-Fi.
- Emphasise the unique capabilities in wide-area connectivity and mobility: the ability to use it on public transport, M2M, etc; emphasise the higher quality of experience afforded by licensed spectrum in an outdoor environment.
- Go for the vulnerable light-user end of the fixed broadband user base with some serious fixed–mobile substitution offers. After all, people will stop using Wi-Fi at home only if they no longer value fixed broadband. This approach would doubtless damage the value of gigabyte, but in the longer run it could help stimulate the release of more spectrum.
The truth is that mobile operators in developed economies are going to have to get used to being the victims, not the perpetrators, of disruptive substitution: fixed and Wi-Fi can do most of what mobile does (except the wide-area/mobility bit) at a fraction of the price to the end user, but mobile can do only a fraction of what fixed and Wi-Fi can because of its inherently limited capacity.
Whatever happens, we believe that mobile operators will have to readjust to higher-than-expected unit prices and lower-than-expected volumes. At worst, they will have to think about managed decline.
Rupert Wood is a principal analyst at Analysys Mason. Rupert's primary areas of specialisation include next-generation networks, long-term industry strategy and forecasting the dynamics of convergence and substitution across fixed and mobile platforms. Rupert has a PhD from the University of Cambridge, where he was a Lecturer before joining Analysys Mason.