Pick your preferred crystal-ball gazer and you'll see forecasts that data use roughly triples every couple of years. That's a particularly big deal in mobile, where spectrum is finite. You can ramp up spectral efficiency via LTE and maximize WiFi-offload, maybe buy some more spectrum, but eventually that data growth translates into building a lot more cell sites, backhaul and core capacity.

It's a challenge that gets tougher over time, which is why most telcos agree that a key pillar of success is getting the economics of data right. Call it "cost-per-bit" if you must, as long as we remember that data use is multimodal -- fixed, mobile, nomadic -- and we need to achieve different costs-per-bit across those different modes to remain competitive and profitable. In fact, to deliver the right economics for data it is going to be critical to drive cross leverage between those different "modes." Whether my tablet is connected via WiFi through the fixed broadband network or via LTE can well make less difference to me (it just works) than to my service provider who is paying increasing attention to the difference in the cost-to-serve.

Indeed, as cell size gets smaller, the mobile network and fixed network start to look architecturally very similar indeed. So similar that it becomes counter-productive to manage them as separate networks. Aside from set-top boxes and servers, most users have a mix of devices that connect wirelessly via cellular technology or WiFi or both. And those devices can be used at home, or at work, or somewhere in between. Strictly speaking, they could be anywhere else on the planet too, but most users' data use mostly occurs in the home-town and so that's the economic challenge to focus on.

What does it take to meet that economic challenge from an organizational perspective? Managing fixed and mobile networks together is part of that equation, but I'd suggest going further. If the economics of data are so important, isn't it necessary to focus a separate organization on that goal? What would such a "DataCo" look like? Would it look like the "NetCos" that are emerging from regulatory pressures or government broadband initiatives?

First let's cover a couple of the structural considerations. We've already said that increasing data intensity and reducing cell sizes accelerates the need to manage fixed and mobile infrastructure as one. But another effect is that it drives an economic need for cooperation between operators serving the same geography. As the number of sites increases, it becomes less and less practical to overlay multiple competing networks. Apart from the raw cost of civil works and backhaul there is also the likelihood that local councils will increasingly resist overbuild as the number of sites grows.

Those factors mean metro infrastructure, especially for small cell and urban WiFi will increasingly need to be shared infrastructure.

The commercial mechanism for that sharing in most cases is probably a simple wholesale model, though joint ownership by two or more service-provider stakeholders is also possible.

Another structural consideration is the importance of "IT infrastructure" in data excellence. Delivering a competitive "customer experience" for data with the right economics is not just about the network -- it's also about optimizing the end-to-end architecture including storage and processing locations.

Obviously service providers are already in the business of caching content and some have much more sophisticated CDN strategies. And some have moved more or less aggressively into the "cloud" space through building, acquiring or partnering with datacenter/hosting organizations. But what is not yet common is an integrated data-infrastructure approach that will serve the service provider's need to excel at "data" across its entire business.

My view is that this is not just a technical issue. It is a fundamental organizational issue -- exceling at "data" means not only managing fixed and mobile together (with an appropriate dose of network sharing) but also managing IT and network infrastructure as one. That's a complex beast to manage well, but that's what needs to be done to get the experience and the economics right. So put that set of assets and management skills into its own organization focused on data excellence. And make sure that that organization is not focused on anything else.

So what does this look like organizationally? As important as "integrated infrastructure" might be, it's not the only thing that matters -- at least as important are managing the overall customer experience and streamlining processes while decommissioning legacy services and systems to deliver operational excellence. While they're quite separate business goals, when we come to actually design systems we find that customer experience and operational excellence are very tightly connected topics because most of the processes and systems involved are customer-centric -- think offers and bundles, subscriber management, contact centers, device management, devices and application ecosystems, customer analytics, and the like. The lion's share of the processes that create the customer experience and that drive the complexity of the business are (not surprisingly) about the customer relationship rather than the infrastructure.

What I'm not saying is that the infrastructure is not an important part of the experience. It will continue to affect brand choice, advocacy and customer lifetime value -- typically more than any other single area! But it is nevertheless one factor among many, and it is not where the majority of business complexity lies. Moreover, data infrastructure requirements are becoming increasingly similar across segments. The trend is clear in devices that there is no longer much distinction between consumer and enterprise. In terms of device preferences and requirements, the differences between any two consumers and between any two enterprises are now much greater than differences between consumers and enterprises in general. And that is increasingly the case for the network -- not that one size fits all, but that the "size" required by a particular user or business is no longer

These considerations favor an organizational model where one organization -- the DataCo -- is focused on integrated infrastructure for data, supporting one or several ServCos that manage the customer relationships, offers, service innovation and go-to-market for the segment(s) they address. The DataCo has all the assets and management talent required to maximize the data experience while optimizing the cost per bit per mode. And it contains only what is needed to achieve that, since everything else is a distraction. So the DataCo would manage datacenter infrastructure, core and aggregation IP networks, fixed broadband, mobile broadband, metro WiFi and small cell, network analytics and network sharing arrangements for data.

To the extent possible, a DataCo wouldn't manage legacy telecom services and it wouldn't manage relationships with customers. Clearly the customer relationship sits with the ServCo(s) and I would argue that the ServCo(s) should also manage the legacy telecom services since the ownership of customer relationships puts them in a good position to manage migration and decommissioning. Of course there can be more than one ServCo for a particular segment -- in the consumer space one ServCo might inherit legacy customers and services with the opportunity to migrate them, while another "challenger" brand works from a clean slate in order to compete with external challengers.

The DataCo can and should consume cable plant and Layer 2 electronics from other providers -- whether for access, metro aggregation or longhaul transmission. But only when it makes sense -- whatever it takes to get the right mix of experience and economics.

Certainly, we cannot pretend that the lines are clean, especially in the case of mobile where towers, backhaul, control plane, and 2G/3G radio are all shared between "data" and legacy services. And no doubt that would be the kind of argument used to remain with the status quo. But the days of that status quo are numbered.

The urgency of change is quite different from one geography to another, but no developed market can escape the need for change in a world where data demands triple every couple of years. So it probably makes sense to think about moving to a DataCo/ServCo model sooner rather than later.

— Philip Carden, founding partner, Number Eight Capital