Re-inventing wholesale voice and data

Telcos worldwide face a key challenge: the decline of their telephony business. One part of this puzzle is the wholesale voice business. A small number of key players deliver telephone calls on behalf of the global industry. A customer of one telco will place an off-network call, and these middlemen will carry that across the world to the destination network for termination.

How can they re-invent this business for a broadband world?

Here is a set of ideas, developed as part of a recent consulting engagement, stripped of all the client-specific references. As usual, a large portion of the credit belongs to my colleagues at Predictable Network Solutions Ltd, with whom I collaborate on such engagements.

Voice is still valuable

Retail users continue to value reliable voice service, and are generally happy to pay for unlimited minute plans on fixed networks. Despite stark declines in some leading markets, demand shows no signs of going away. The take-up of substitutive consumer Internet voice (e.g. Vonage and Skype) has been sluggish. This all suggests that consumers and small businesses value dependable, packaged services supported by a single service vendor.

New incremental forms of demand for IP voice, driven by technologies such as WebRTC and Hypervoice, appear to be due for significant growth in the next 3-5 years. Furthermore, users are adopting an increasing range of services with “voice-like” quality needs, such as 2-way video, unified communications, femtocells (where voice is one its constituent services) and gaming. These benefit from the kind of data transport quality previously associated with constant bitrate voice. This has traditionally been delivered using reliable circuit-based networks based on time division multiplexing (TDM).

The nature of supply, meanwhile, is also changing. IP voice is slowly mitigating the underlying issues and barriers that stand in way of its substitutive use for TDM. It is slowly diffusing into the main voice market, and an all-IP PSTN is being discussed. Furthermore, there are market context changes, such as a potential move to a bill-and-keep regime (where no money changes hands for terminating the call).

The nature of the wholesale voice business today is relatively simple: to carry flows either over IP with an implicit quality need, or with TDM switching and an explicit quality need. To meet the quality need, the service is sized to peak capacity requirements. The industry has a special term to define the sizing parameter: PSTN voice erlangs. In this voice world, there is a clear outcome – a quality call – and a variety of well-established measurement methods such as mean opinion scores (MOS).

A whole market of measurement of the voice quality outcomes has sprung into existence as part of the assurance process. There is an existing body of work on how MOS and the technical data transport measures interact. Using this to underpin any technology or service transformation helps to mitigate and manage technical risk.

Challenges in continuing the current path

That voice business is coming under increasing pressure. Prices are falling, and buyer power is increasing as the retail business is consolidating. Both top revenue and bottom line profit are expected to fall.

Specifically, the incentive structure of the market is going to shift with the end of metered telephony:

  • There will be less need for the kind of complex network operations and billing systems currently used to optimise routes and cost. This lowers barriers to entry.
  • There will be new challenges in maintaining call quality as players in the market have perverse incentives, and may no longer endeavour to deliver good call quality. We already see this in the USA with rural call quality, as calls get least-cost routed so much that the induced latency makes them useless. That means operators will need an unambiguous approach to allocating responsibility for any service quality issues, and to monitor and trend changes.
  • ‘Over the top’ services will continue to grow and erode volumes and perceived value.

There is also technological change. We are going from space multiplexing (circuits) to space-time multiplexing (packets). This is creating residual risks for the service deliver, which are quantifiable, but will be seen as novel to the existing skill base and processes.

Cannibalisation of existing revenues may be required in order to build the kind of wholesale business that is likely to be viable in the future. For example, to create scale in growth areas in a bill-and-keep world, it may be necessary to price those products at levels that undermine the existing metered minute model. For this to be palatable to management and investors there has to be a compelling vision and business case.

Avoiding the “death spiral”

The situation that every wholesale voice network operator wants to avoid is carrying a high fixed cost structure, for a declining product, with a need for significant IT investment to meet regulatory change, and with quality issues over which it has little control. Worse, in this market structure IP voice interconnect starts to be delivered via the Internet, rather than over a separate managed voice network. The un-assured IP delivery approach creates a new hazard, since as quantity increases, quality declines. The pressure for cost reduction from the CFO begins to arm and activate quality of experience (QoE) hazards in the network which becomes ‘over hot’. This is not just a hypothetical situation: we see this in our consulting work.

In this worst-case future scenario, there is no synergy between a voice business and any other lines of business (e.g. content delivery networks and data services). Many wholesale voice operators will be forced to exit the wholesale voice market, causing an oligopolistic market concentration among those who remain. Meanwhile, the adjacent enterprise voice market becomes over-crowded with many entrants, and has powerful application intermediaries who capture most of the profit pool.

The retail broadband market becomes a one-way trap, whereby revenues have fallen below a sustainable level, but prices can’t be raised. Users feel an entitlement to communicate, but do not see any new service offering or value that would justify raising prices. In this scenario, no new markets for IP wholesale products emerge.

Building a successful converged wholesale business

Instead, what a wholesale operator wants is a business which has economies of both scale and scope, and delivers a wide range of services at price points that cannot be replicated by its competitors. The best outcome is one where it has a virtuous circle of more user volume and value driving more services onto its network platform, which then increases the number of users. In this model, the operator can offer services tailored to multiple price and quality points, and service a wide range of types of application with wholesale transport needs. There is strong synergy with other lines of business. The business has strong differentiation and barriers to entry.

The issue to be solved

How can a wholesale voice operator re-position its business to maintain scale and reduce costs in its legacy cash-cow business, which is being transitioned from a circuit to a packet model, whilst enabling new differentiated wholesale products?

How to build a converged quality-assured voice and data wholesaler

To achieve this, you need to align with and exploit the fundamental properties of packet multiplexing. This enables the cost structure and delivery capabilities that will be successful in future.

For the legacy circuit voice market, the operator must construct a voice data transport delivery system which has observational equivalence with circuit voice, but does not carry that cost structure. This equivalence is in terms of delivered application outcomes, i.e. MOS and an equivalent metric for signalling. The technical requirements of voice delivery, in moving to IP based transport, by necessity have become relaxed.

Setting pure TDM technical equivalence as a goal would be a big mistake: it is the user outcome, rather than the technical input, that matters. What we are producing is a product best described as “I can’t believe it’s not TDM!”.

That means both increasing how ‘hot’ the network is run, as well as using new techniques to transform cheaper generic connectivity inputs into quality-assured voice transit.

For the new growth application areas, the operator must keep the value high by continuing to sell application outcomes (as is the case now with circuit voice), rather than undifferentiated bandwidth inputs. Selling bandwidth merely allows other parties in the service chain to capture the value.

This can be achieved by creating an “assured service erlang” that goes beyond voice telephony. The differentiation is fitness for-purpose: you enable customers to depend on outcomes, and act as critical national/business infrastructure for vital applications. What was once true for a single application (voice) becomes a generic quality assurance market for all cloud services.

From assuring voice to assuring all cloud services

We take what made voice successful, and abstract from that the transferable technologies and business processes. We then turn those into a number of specific services like “video call erlangs” or “femtocell erlangs”. The wholesale operator delivers service-oriented outcomes with a measured unit of worth, such as completed video calls.

In this model, the wholesaler can also be a market-maker. This is a little like how Amazon is both a retailer and platform provider. This model allows other parties to create and own the physical assets, while the wholesaler composes them, manages the technical and contractual interfaces, and collects the value of creating end-to-end value for the customer. Furthermore, the wholesale operator can be an aggregator of relationships with retail ISPs for assured delivery, a little like how the CDN market works today.

The advantage of this model is that he who sells the services gets to control the multiplexing. You cannot be bypassed, and you will accrue the full benefit of having assured it. Additionally, there is a natural follow-on sale of local computation and storage, since the best cost and experience outcomes will be on your assured network.

To keep up to date with the latest fresh thinking on telecommunication, please sign up for the Geddes newsletter