These days all the attention in the US wireless industry seems to be on the growth of data traffic. And yet, despite all the growth, it is voice that US mobile subscribers seem to love more than most in the world--and comparatively consume the same if not less data. AT&T subscribers talk for 622 minutes on average, compared to 136 at NTT DoCoMo in Japan, and 162 in the U.K. (average across operators, according to the U.K. regulator Ofcom). In a recent survey among WiMAX operators, I found to my surprise that the heaviest data users were not in the U.S. or Japan (7 GB average per month at Clearwire and UQ), but in Russia (13 GB) and followed by South East Asian and Middle Eastern countries.
The second data crossover
The first voice and data crossover was in traffic. By 2009, Vodafone data traffic was 178 percent of voice traffic, with voice traffic essentially constant through time, as the number of subscribers and their usage levels reached saturation. The growth in data traffic has been faster than the growth in revenues, as subscribers increasingly move towards data intensive applications (e.g., from texting to video), but they price-per-bit quickly drops. Texting has been for a long time the most profitable service for mobile operators (even more so than voice), but with flat-fee rates, and larger data allowances or unlimited plans, the profitability of data services has for many operators decreased as SMS has ceased to be the main source of data revenue.
Despite all the growth in data ARPUs, data still accounts for a smaller portion of revenues than voice, despite using the majority of network resources. In the US, data accounts for only 35 percent of ARPU at AT&T, and 36 percent at Verizon, even though about a third of subscribers have a smartphone. At Vodafone in Europe 30 percent of ARPU comes from data; in Africa 19 percent; in India 11 percent. At the current growth rate, the balance between voice and data is set to reverse soon.
Japan leading the way
In Japan this is happening already at Softbank and NTT DoCoMo. At NTT DoCoMo voice and data ARPU are pretty much the same, but by the end of the current fiscal year (March 2011 in Japan), data revenues will be higher than voice (Figure 2), driven by the combination of lower voice use, and growing adoption of more intensive--and expensive--data plans, and of video applications (Figure 3). At Softbank, pushed by the exclusive and very successful deal on the iPhone, the data ARPU (¥2030) has crossed voice ARPU (¥2250) in the last reported quarter.
The voice/data ARPU reversal marks a major shift in the wireless industry. Voice will see its role of cash cow for mobile operators reduced, not because as some feared it would be replaced by VoIP, but because subscribers use increasingly their phones for data instead of talking to each other. As expected, this change in behavior is especially notable among younger subscribers, as data from the UK indicates (Figure 4).
And what about the Philippines?
For a long time, the Philippines have been the country with the highest percentage of data ARPU--it is also approaching the 50 percent mark. The situation there is somewhat different. The relatively high data ARPU is due to very low data revenues and a subscriber base that loves texting. The opposite of the US, the Philippines are a market where data has been the main growth driver for a long time and, despite the good start, the relative growth of data ARPU proceeds more slowly.
It is not surprising to see that this is happening in Japan. Data usage and corresponding ARPU has been among the highest in Japan since mobile data got off the ground. In light of the most trite sociological stereotypes, it is not surprising that loud Americans prefer to chat endlessly on the phone, while the more reserved Japanese prefer means of communications that are far less obtrusive to the people around them. And if you spend, as most people in the US do, one or two hours alone in your car driving to work, you have plenty of time to get your MOUs up to 1000 minute mark (by comparison, at NTT DoCoMo voice plans start with an allowance of 25 minutes, which is what many people use in one day in the US). (Unfortunately the most resourceful among us in the US have learned that texting too can be used to relieve the boredom of the commute, but that's a topic for another day). If instead of driving, you spend the same time in public transportation or dense urban environments, you are less likely to find it pleasurable to talk over the phone for long times.
More than stereotypes
Yet the difference in voice and data usage is sufficiently large to suggest that there is something else at play that makes the US stand out as the exception--and not only with respect to Japan. The high MOU levels in the US have been triggered and sustained by an extremely favorable pricing for voice. To drive adoption of the service, US operators offered generous buckets of minutes to their subscribers, while operators in other countries used metered billing for voice calls. With a monthly allowance, subscribers try to maximize the time spent on the phone before reaching the limit; with metered charges, subscribers try to minimize the amount spent on the phone, to the advantage of services like texting that sharply reduce charges. For a long time, for US subscribers data was more expensive than voice simply because it was not included in the monthly allowance--and it is still is not in many cases.
In Japan, both data and voice have been metered for a long time, and as a result subscribers have not been steered towards voice by cost considerations. As operators have moved towards a more diverse set of service plan options that include both metered and flat-fee, often the monthly allowance can be used towards any mix of voice and data services. The equal treatment of voice and data, as well the closer integration within the two types of services is likely to have encouraged the transition to data, to the point that voice usage has become threatened by data. In the US, the same process is under way, with MOU declining over time and data increasingly compensating in terms of usage and revenues.
The difference of course is that that gap between voice and data was much larger at the beginning in the US and operators did not remove the incentives to use voice--and this slowed down data growth in the initial stages. However, the path is clear even for chatty markets like the US. It is only a question of time when data revenues in the US--or Europe--will reach voice revenues. In fact the very distinction between data and voice is bound to disappear or lose most of its significance as voice increasingly gets treated as a data application.