I used to think that in the mobile world it is voice that subsidize data. The rationale seemed obvious: voice requires little bandwidth, but generates the largest proportion of revenues. Data, and the other hand, uses a disproportionate amount of network resources and has little to show in terms of revenues.
This might have been the case a few years ago, when data services where an optional, ancillary service that subscribers used parsimoniously, and with limited usefulness. Today, the balance between data and voice has fundamentally changed. Voice no longer subsidizes data, and for an increasing number of subscribers the two are equally essential components of their mobile service.
While is clearly a case of stating the obvious, the evolution in the voice/data relationship has some deep implications on the relevance in usefulness of data ARPU as the metric to understand operators' performance and their ability to generate revenues from data services. When looking at per-bit revenues, the distinction between total ARPU and data ARPU can be downright misleading. We are so used to looking at data ARPUs and revenue figures, that we do not realize how anachronistic they are.
The good news is that some operators have noticed. Verizon, for instance, has stopped quoting data ARPUs as a key metric, and some European operators have done the same. Still, it is common to see the total versus data split, even in countries like Japan where data revenues have surpassed voice revenues.
What makes data ARPUs dangerous or, the best case, irrelevant? After all, they are the bright spot in service revenues. Voice ARPU is declining, data ARPU is growing.
However, when you compare at the revenues per gigabyte for voice and data, it becomes clear that data ARPU is no longer an appropriate metric. As voice accounts for a decreasing percentage of overall network traffic (7 percent in the Vodafone's network in Europe), its profitability continues to look fabulous. On the other hand, the business case for data is dreadful: little marginal revenues, lots of traffic.
We can look at Vodafone in Germany for an example. I choose Vodafone because it publishes more comprehensive financial and traffic data, but I expect the results to be similar for other operators. Monthly ARPU is about $20, with data accounting for 34 percent of revenues ($7.50), according to Vodafone. German subscribers use 150 minutes of voice calls per month and 280 MB of data – i.e., a total of 300 MB when voice is included. The per-MB revenues for combined data and voice are $0.08 (calculated from the $22 total ARPU), while for data only they are $0.03 (calculated from the $7.50 data ARPU).
Does this interpretation of mobile network performance correctly capture the financial dynamics at play--or even the perception among subscribers? I would argue it does not. If that were the case, you would see mobile operators offering exclusively voice services and see their profitability go up compared to their peers stuck with data-hungry subscribers with iPhones and Android smartphones. Subscribers value highly both voice and data services and would not settle for one at the exclusion of the other. I suspect that for some of us data is more important than voice. It certainly is for me.
As a result, to compute data ARPU as the marginal revenues subtracted from a total ARPU that includes also base plan and voice revenues underestimates the true weight of data revenues. At the minimum, part of the base plan revenues should be allocated to data.
This would be in line with what is happening already. Many mobile operators charge their smartphone subscribers for a plan that includes a mix of voice minutes and data megabytes, understanding that the distinction between the two services would only unnecessarily increase the complexity of the plan. In Germany, 62 percent of Vodafone's revenues are now coming from such integrated service plans, according to the operator. Furthermore, services such as voice and SMSs that generate little traffic are now often thrown in high-end plans as unlimited. At the same time, operators have introduced stricter and more actively enforced data quotas, because this is where the cost to provide the service come from. Over the next years, the transition to VoLTE will make the separation of voice and data services--and hence revenues--even harder to make, if not impossible. More importantly, it will be meaningless.
What is the alternative then? First, we should wean ourselves from our reliance on data revenues or ARPUs, as they are unreliable indicators of the contribution of data services to profitability. We need a way to capture the profitability of both voice and data services, and assess the ability of mobile operators to generate revenues from their networks. A simple and yet powerful way to do so is to publish the revenues per MB--or, to be futureproof, per GB--as this will make it possible to evaluate the ability of mobile operators to extract value from their network assets and the return on their investments. It is possible to compute revenues per MB already as we did with that data from Vodafone, using publicly available data, but estimates would be more consistent if operators routinely published them as one of the key metrics.
Monica Paolini, PhD, is the founder and president of Senza Fili Consultingand can be contacted at [email protected]. Senza Fili Consulting is an analyst and consulting firm that provides advisory services on wireless data technologies and services.