Beyond bundling: the future of broadband

This is an edited version of the keynote presentation of Martin Geddes, Chief Analyst at STL Partners, at the Telco 2.0 Executive Brainstorm in London last month. It provides some initial findings from our research into future business models for broadband service providers (BSPs), including our recent online survey. (The summary results will be mailed out to respondents in the next few days.) Those wishing to find out more may want to take a look at our forthcoming report, Broadband Business Models 2.0.

To save you the suspense, here's the headlines for what's upcoming for the telecoms industry, based on what insiders are saying through our survey and research:

  1. Operators are going to face a slew of non-traditional voice service competition. To corrupt the words of Yogi Berra, "The phone network? Nobody goes there anymore, it's too crowded." The volume may linger on, but the margins in personal communication will move elsewhere.
  2. Content delivery is a logistics problem that spans many distribution systems. Those who can solve the delivery problem by sewing together many delivery services, rather than those focused on owning and controlling one channel, will win.
  3. Wholesale markets in telecoms are immature and need to evolve to support new business models.
  4. Investors aren't up for more "loser takes nothing" facilities-based competition capex splurges. Time to look hard at network sharing models.

So, read on for the background and evidence:

Background to the survey and research

Our ingoing hypothesis is that telecoms - fixed or mobile -- is a freight business for valuable bits. This could be via traditional voice networks. Broadband is another means of delivering those bits. It includes Internet ISP access, as well as other services such as private VPNs and IPTV.

Broadband competes with and complements other delivery systems like broadcast TV, circuit-switched phone calls and physical media.

Just as with physical goods, there are lots of delivery systems for information goods. These are based on the bulk, value and urgency of the product - from bicycle couriers to container lorries for atoms; phone calls to broadcast TV for bits.

As part of our research we've also been looking at how other communications and delivery systems have evolved commercially, and what the lessons are for the telecoms industry. After all, broadband as a mass-market business is barely a decade old, so we can expect considerable future change. In particular, the container industry has some strong parallels that may hold important lessons.

Physical goods and the telephone system have developed a wide range of payment methods and business models.

With physical goods we have "collect it yourself", cash-on-delivery, pre-paid envelopes and packages, as well as express parcels, first and second class postage.

The phone system offers freephone, national, non-geographic and various premium-rate billing features. It offers the user a simple, packaged service that includes connectivity, value-added features, interoperability, support and a wide choice of devices.

Likewise, SMS packages together the service and its transport. It's wildly popular, bringing in more money globally than games software, music and movies combined.

The problem is that this has come within closed systems that don't enjoy the rich innovation that the open Internet brings.

Internet access, by contrast, offers an abundance of goods but is relatively immature in the commercial models on offer. Broadband service providers typically offer just one product: Internet access. And they generally only offers one payment mechanism for delivery of those online applications: one-size-fits-all metered or unlimited, paid independently of services used. (There are some important exceptions -- you can read more here.)

As a small example of how the Internet under-serves its users, when a small non-commercial website suddenly gets a surge of traffic it typically falls over and is swamped. That is because there's no commercial incentive for everyone to pay for a massively scalable hosting plan just in case of unexpected demand. The telephony system doesn't suffer this because the termination fee for every call is designed to at least cover the technical cost of carrying the call.

Oh, and don't expect Google to host it all for free for you either - the error message in the slide above is cut and pasted from a bandwidth-exceeded Google Blogger account.

There is also a lack of incentive for access providers to invest in capacity on behalf of Google to deliver richer, heavier content (where Google collects the revenues).

The question therefore is: How can BSPs find new business models inspired by more mature distribution systems?... whilst at the same time not killing off the innovation commons that is the Internet. BSPs must both create and capture new value in the delivery of online applications and content. Being an NGN or IPTV gatekeeper is not enough.

Fixed voice revenues are declining; mobile voice is peaking; and SMS is slowing down. The theory has always been that broadband ISP services will take up the slack, but in practise margins are thin.

Our research is testing out a wide variety of alternative commercial models. For example, would an advertiser like Google pay for not just the hosting of content (via YouTube, Picassa or Blogger), but also the end-user usage on a fixed or mobile device for receiving that content?

We believe that whilst these alternative models may individually be much smaller than traditional broadband Internet access, collectively they may add up to a larger amount of value.

Survey supporters and respondents

The research would not be possible without the active support of the above sponsoring and supporting organisations, and we thank them all.

We've had over 800 respondents, with roughly one third from operators & ISPs; a quarter from vendors; and the rest consultants, analysts, etc. The geographic split is Europe 40%, N America 30%, Emerging 20%, Developed Asia 10%. There is a ratio of around 60:40 fixed:mobile respondents, and mostly people from commercial (rather than technical) functions.

We asked about four main areas:

  • Today's ISP model -- is it sustainable.
  • Future of voice service in a broadband world
  • Future of video service, as the other leg of the "triple play" stool
  • Future business and distribution models

Rather than assault you with dozens of charts and statistical analyses, what follows is the gist of what we've discovered.

Furthermore, we're looking 5-10 years out at macro trends. You might not be able to predict Google, Skype or Facebook; but you can foretell the rise of search, VoIP and socially-enhanced online services. Even in our own industry, there can be large structural changes, such as the creation of Openreach by BT. You could probably have foretold that as vertical integration weakens there would be such organisational upheavals, even if not who and when.

Sustainability of ISP business model

What's the future business model for broadband?

Around 20% see the current stand-alone ISP business model as sustainable long-term. This includes many senior industry figures, who cite better segmentation, tiered price plans, cost-cutting and reduced competition in more consolidated markets. It may be a minority view, but cannot be dismissed out of hand.

Around a quarter of respondents thought that broadband works as part of a triple or quad-play bundle of voice, video and data - cross-subsidised by its higher-margin cousins. This is the current received wisdom.

However, a majority of respondents say that a new business model is required. These results hold broadly true across fixed and mobile; geographies and sectors.

Which brings us to our first lesson from the container industry. Old product and pricing structures die hard. The equivalent efforts at maintaining a "voice premium" all failed. Trying to price traffic according to the value of what's inside the container or packet doesn't scale.

For BSPs, that means technologies like deep packet inspection might be used:

  • for law enforcement ("x-ray the containers"), or
  • to improve user experience (at the user's request), for example by prioritising latency-sensitive traffic ("perishable goods")

However, traffic shaping can't be your only or main tool for the long-term; you can't reverse-engineer a new business model onto the old structures. It doesn't, ultimately, contain your costs or generate significant new revenues.

Broadband voice

One of the big surprises of the survey was how quickly respondents see alternative voice networks getting traction. We asked what proportion of voice minutes (volume - not value) will go over four different kinds of telephony in 5 and 10 years from now. Looking at just the growth areas of IP (i.e. non-circuit) voice, you get the following result.

It seems those WiFi phones we laugh at now are more dangerous than previously thought - maybe when 90% of your young customers are communicating via social networking sites, you've got some unexpected competition? (Indeed, we note that social network traffic is just overtaking the traditional email portals.)

We were also given a surprise in that respondents saw most of these changes happening over the next 5 years.

Insiders see the growth in voice traffic as being anchored on best-effort Internet delivery, which gets around 1/3 of the IP voice traffic. Using traffic shaping, offering tiered levels of priority, and using traditional end-to-end quality of service guarantees all got roughly equal share.

There are some small differences between fixed and mobile, and mobile operators might like to seriously consider offering tiered "fast dumb pipe" and "slow dumb pipe" that applications can intelligently choose between.

This all suggests that operators may be over-investing in complex NGN voice networks and services. They need to urgently work out how they can partner with Internet application providers to offer "voice ready" IP connectivity without the costly telco-specific baggage of telco protocols and platforms.

So what's the lesson from container shipping for the broadband voice community?

At the same time as containers where being adopted, some ports doubled-down on the old business model and built better breakbulk facilities - and lost. Manhattan's quays are gone, Newark has replaced it.

Others waited to become "fast followers", and lost too. London went from being one of the world's busiest ports, to zero activity. Dubai did the reverse by investing exclusively in the new model, with a low cost base and high volume. (Shades of Iliad's approach in France.)

The winners were those who staked out the key nodes of the new value chain.

There are some clear lessons here for telcos and their NGN voice networks. The cost of broadband access technology is dropping, capacity is rising, and the voice component's value is decaying towards zero. Furthermore, session control (the software part of the voice application) is just another IT function that runs inside a big server, and isn't something you can charge for above hosting costs. It has the economics of email, and that's mostly given away for free. So IP voice isn't adding anything to your triple/quad play bundle, and can only be justified on the basis of reducing cost in the old business model. An IP NGN voice service that's still selling metered minutes does not constitute a new business model.

Broadband video

The survey results for video are a little less dramatic than for voice and follows received wisdom more closely. Overall respondents endorsed Internet video as far more of an opportunity than a threat. (Only in telecoms can a significant proportion see more demand for their product as a problem! The potential issue is that video could drive up costs without sufficient compensating revenue.) A long slow decline for broadcast TV and DVDs is matched to a slow ramp-up in various forms of on-line delivery. Every form of Internet delivery, from multicast IP to peer-to-peer file sharing gets a roughly equal cut. There were some things to watch out for though...

The opportunity is to become as supplier of advertising, e-commerce, caching and delivery services for a variety of video portals, not just tied to your own. This isn't surprising; can you imagine a Web where there were only two portals to choose from, both owned by the network owners? The same applies to video.

Economic migration, cultural fragmentation and user-created content ensure that we'll need a diversity of aggregation, recommendation, filtering and presentation technologies.

Given a choice between building a closed IPTV solution, or an open content platform, the response was well in favour of the latter as the more profitable to run. (The slow ramp up of BT's Vision service suggests its success is more likely to be based on the "push" of analogue switch-off than the "pull" of the telco brand as a TV provider. Why do no telco TV plans centre around external entrepreneurial talent and innovation?)

Both options beat the alternative of disinvestment in video delivery technology. So fixed and mobile operators are well positioned to help enable and market video, just not "TV over IP". That's the steam-hauled canal boat, when you're supposed to be using IP to build a railroad. It seems telcos are over-investing in emulating broadcast TV and under-investing in the unique nature of the online medium.

P2P and "over the top" are here to stay. You deal with the costs by offering more profitable alternatives, not by punishing your most voracious customers. (See our article on Playlouder as an example of how to do it right.)

In music, Apple's iTunes captured the key bottleneck in the distribution chain. Could the same happen for online video?

We gave respondents a choice of four scenarios:

  • Direct to user from the content author or publisher
  • A single dominant player
  • A fragmented market dominated by telecoms companies
  • A fragmented market dominated by non-telcos

Our respondents say that the market is likely to be fragmented with many aggregators and non-carriers will dominate. Again, "triple play" doesn't capture the richness of the business-to-business model required with many partners in the distribution and retail value chain. How will Telco TV satisfy my wife's taste in Lithuanian current affairs and my interest in gadgets and economics lectures? It can't.

Our take-away from the shipping industry is that when it comes to shifting bulky stuff around, big is good and bigger is, err, gooder. Networked infrastructure businesses have strong increasing returns to scale. There's no point in building a new port anywhere near Rotterdam because that's not where the other ships go. There's a good reason why Akamai takes the bulk of the profit pool from content delivery networks -- their one is the biggest.

Network ownership models

Compared to today's dominant models (facilities-based competition and structural separation), respondents rated a third ownership model - co-operatives of telcos - surprisingly highly. The two currently dominant models remain on top.

The issue is how to structure the vehicles for mutual or co-operative asset ownership. The financial industry has already created structures that allow shared operational businesses, either mutually owned or as private special entities. Furthermore, they've managed to preserve barriers to entry. To become a member of the VISA network, you need a banking license. That costs a lot of money.

Telecoms and the Internet business have some common structures around numbering and interconnect, but could emulate these other models from other industries.

The arrival of containers shifted the balance of profit away from the shipping lines and towards the ports.

In terms of telecoms, it's where the content is originated or goes between delivery systems that matters - from CDN to broadband access, from broadcast to DVR. That means every Googleplex and content delivery network that gets built puts Google or Akamai at a massive advantage, since everyone wants to peer with them.

Traditionally it has been long distance and access networks that have dominated telecoms economics. AT&T's early years found it the only owner of a long-distance network and thus able to negotiate very advantageous terms in buying up local carriers into the Bell system. It mistakenly help onto the long distance network just as the bottleneck shifted to the access network. At the moment the US sees a duopoly in access networks, and supernormal profits. Wireless carriers enjoy an oligopoly in most markets as a by-product of spectrum licensing.

However, Europe is moving towards structural separation or open access of fixed networks. Homes and offices offer WiFi or femtocell bypass options for cellular. Over time, local access ceases to be such a bottleneck. Furthermore, there are many physical paths and proliferating technologies and suppliers hauling data between the distant points that want to be connected up -- be it transoceanic cables or competing wireless backhaul technologies. So the owners of the transmission networks don't enjoy the benefits. It's the owners of the places where traffic is exchanged between delivery systems that do, since those feature increasing returns to scale and dominant suppliers.

What is the product we are selling?

Today operators expect you to go out and buy yet another access plan for every device you touch or place you make your temporary home. They sell "lines", either physical, or virtual (via a SIM card). Is this really the right way for the future?

All I want to do is connect my phone and laptop to the Internet wherever I am - but I get different prices and plans depending on which combination of device and access technologies I use - yet all from a single vendor. (The first is using my phone as a 3G modem over a USB cable; second is a separate 3G USB modem; third is WiFi.) This creates the perverse incentive when I'm sat in Starbucks to use my phone as a modem for my laptop over the expensive 3G network.

Also, I might be a peer-to-peer download lover, and hopelessly unprofitable. Or I might just want to check my email and surf the web a little on my mobile. How can you rationally price this product? What are the alternatives?

We gave users a choice of 3 alternatives (above) as to how broadband connectivity is provisioned. Should we sell you "unlimited browsing", but listening to Internet radio is a separate charge? Or should we price access according to the device, but not make the plan portable between devices? A data plan on a basic featurephone would differ in price from a smartphone, Internet tablet or laptop. Or should we just give the user a set of credentials that activates any device or network they touch and bills that usage back to them?

The preferred one was to offer users a connected lifestyle, regardless of devices, applications or prices.

BT's deal with FON is an example of a step towards this goal. Picocells too have the potential to upend the access line model. In terms of immediate actions, mobile operators should recognise the trend towards divergence and users with multiple handsets. Don't make me swap SIMs around when I go from my "day phone" to "out on the town phone". Give them a common number and interface.

New, more liquid, ways of combining together devices and networks for sale would require wholesale markets to evolve.

We asked what impact it would have on BSP revenues if all the friction were taken out of the wholesale market. Anyone who wants to come along and build an application with connectivity included in the price would be able to source their wholesale data from any carrier. You don't have to be Yahoo!, Google or RIM to negotiate a deal with every carrier in the world, or make one-off special billing integration.

The effect? A 50%+ boost in revenues, which has a commensurately greater effect on profit. How much value is the broadband industry leaving on the table because of its inability to package up and sell its product via multiple channels?

Even more profitable than the ports are the agents who arrange the end-to-end logistics and supply chains for their customers. In telecoms terms, it's the operator who can assemble a multitude of fixed and mobile networks, content delivery systems and B2B parterships with the application providers that wins.

For telcos, the critical development to enable personalised packaging of connectivity, applications and devices is to build richer wholesale models. The hot activity will be in the B2B markets, not direct-to-user. The failure of most MVNOs has shown that you don't just want to create "mini me" telcos, but to enable more granular offerings.

Conclusions and summary

Telecoms is going to move to a multi-sided business model. Google are as likely to be paying for the full delivery of the ad-supported YouTube video as the user is. The telco will also feed Google usage and relationship data to help target advertising. Google might use credit data from the operator to manage its own fraud and chargeback risk on its checkout product. Telcos are logistics companies for data, helping the right data to be at the right place at the right time. This is completely different from being a "dumb pipe", wannabe media company or end-user services provider.

When you buy a new electronic gizmo, it typically comes with batteries included. The battery makers have learnt to supply batteries wholesale to consumer electronics makers, as well as to end users. Broadband needs to evolve to add "connectivity included", with the right quality and quantity packaged up with the application or content in ways that the user finds easy to buy. Today's product is selling users a raw unprocessed commodity, which is serving neither the interests of the users, merchants or operators.