Summary: As O2 UK’s Litmus developer proramme matures into a global corporate project for Telefonia, we analyse the business model challenges it faces in becoming a vibrant community for developers and a value driver ofr the company.

Back in March, we said that O2’s Litmus developer site was “better than the Apple App Store”. Quite a claim, as it turned out. We based it on the deep integration of Litmus with the range of social and business enablers it provided in addition to the O2 network APIs. As well as a generous revenue share and quick payment, Litmus offered access to O2’s billing system to help cash collection, crowdsourced testing from Mob4Hire, Web-hosting services, and the tantalising prospect of access to an internal Telefonica venture capital group.

How is Litmus doing now?

In terms of product quality, Litmus’s recently added some highly interesting APIs. For example: the ability to query the current status and capabilities of a device, whether the user has sufficient credit to make a payment, if they have an inclusive data plan, whether they are in a WLAN hotspot, and whether or not they are currently roaming.

The importance of this kind of contextual data – call it Level 1 context – for delivering an excellent user experience with mobile applications and content is hard to overestimate, and it avoids most of the political issues that dog some other forms of context, like user behaviour and social graph data (call them Level 2 context). Overall then, the potential quality of application looks encouraging.

But how about quantity? At the moment, there are 36 pages of apps on sale at Litmus, plus three more for testing; at 10 apps to the page, that’s 390 apps. Many of them are versions of the same application for different devices or localisations, so the count of active projects is rather less than that. It’s also true that a lot of people submit their applications to every app store going, sensibly enough, so there is quite a bit of duplication.

So far, this is a respectable try, but it’s nowhere near Apple’s app count. However, as we’ll see later on, stacking up apps in an app store isn’t the only strategy available.

A further indicator on the quantity of development activity is that the forums at o2litmus.co.uk look worryingly quiet. Another traditional measure of activity at an open-source project is the traffic on the mailing list; there doesn’t seem to be that much going on. This is something Litmus has in common with the other mobile dev platforms – the Symbian and Forum Nokia ones are patchy at best. Perhaps this point from The Information Architecture of Social Experience Design‘s list of anti-patterns for Web sites applies:

“a Potemkin Village is an overly elaborated set of empty community discussion areas or other collaborative spaces, created in anticipation of a thriving population rather than grown organically in response to their needs”.

So, why aren’t we seeing much more development activity at Litmus? It’s a big question, especially as Litmus is meant to be under active development. What are the warning signs of a community that might end up looking like this?

The critical challenge is getting to sufficient scale, which is vitally important to the success of platform business models like Litmus. O2 UK has 18 million subscribers; if 10% are conscious of apps, that is an addressable user base of c1.8 million.

Further, it’s probably true that iPhone owners tend to be power users, being a self-selected group of early adopters. (According to Ray da Silva of Vodafone, iPhone users exhibit 7 times greater usage than the closest rival group, BlackBerry users.) And O2 has the exclusive right to distribute iPhones in the UK, so the bulk of O2’s power users are probably concentrated in its population of iPhones. Those 1.5 million O2 iPhone users have the App Store to go to, which is integrated with the hardware and software and prominently placed on the device. If our estimates are close, that leaves about a fifth of that number, or 300 thousand or so who might use Litmus.

So, Telefonica / O2 faces a strategic dilemma. How should it balance investment in creating and serving the huge (but ultimately Apple’s) iPhone community and the nascent and home-grown Litmus eco-system?

And, as we’ve often pointed out, telcos consistently overestimate the degree to which their subscribers constitute a real community or want to have any affinity with their operator. Apple, at least, can claim to be the proud owner of a cult, an image it works extremely hard to maintain. Probably no other hardware vendor in mobile can claim that, and the OS vendors aren’t much better off although Symbian tries hard.

This is important, because active developer communities tend to be driven by a smallish core group of members. Recruiting new members of this group is critical for long term survival. On the other hand, the problems, ideas, feedback, and money coming into such a community usually originate in another community core group – the user elite. The line between the power users and the developer community is necessarily fuzzy, but it’s crucial that you have enough people in the user community who are passionately engaged with the product to support the developer core group.

Fragmentation is another challenge resulting from insufficient scale; it’s a serious problem if you have to keep refactoring your code to work on dozens of different devices and OS platforms. Equally, being fragmented between operators is no better; in terms of scale, developing for Symbian is going to beat developing for O2 UK.

Put together, these issues add up to a serious overall challenge to the viability of Litmus in its current form as anything other than a test of limited scale and ultimately limited value.

Litmus Responds…

So, clearly it was going to be interesting when James Parton and Jose Valles Nunez, from Litmus and Telefonica’s Open Innovation group respectively, dropped into the Telco 2.0 offices.

The first interesting point that arises is that the Litmus group within Telefonica is very keen not to be considered an appstore. You might think this is a brave decision; everyone in the industry is obsessed with them since Apple’s big hit, and a week doesn’t go by without someone launching one – whether an operator, a vendor, a third-party store like Handango or Symbian’s app warehouse, or a gaggle of hackers doing an unofficial one for iPhone apps that Apple don’t like.

The obvious corollary to that is “well, what is it then?” Parton argues that the real role of Litmus isn’t as a first-line product, but rather as a way of crowdsourcing decisions about which applications to promote to the mass market through O2 Active – a form of “co-creation” with the community of power users and developers. Rather than relying on the judgment of product managers in Slough, the idea is to serve up new ideas to a self-selected group of neophiles and to see what sticks. Litmus is hoping that this will both provide useful feedback and also reduce churn by binding their user elite into the company more closely.

So far, they report that the extra features like hosting and testing haven’t been much used, and were perhaps a case of “over-engineering” the product – most of the developers involved are primarily interested in Litmus as a route to market, whether as an app store or as a sort of X-Factor for applications that might make it to the official O2 deck. However, they are keenly concerned about recruiting more developers and about the perception of a lack of critical scale.

Scope for Business Model Innovation

So perhaps Telefonica, and the industry as a whole, should be looking for other organising principles for developer communities – whether to build scale in their own right or just to get to ‘critical mass’ in the communities? Rather than being operator- or vendor- specific, perhaps they should be application-specific or problem-specific?

The main forces that create these communities are either technological opportunity – ‘we can do something new!’ – or else an urgent problem – ‘how can we fix this?’ Examples of opportunity-based communities include the vigorous groups that grew up around major programming languages, or the Linux kernel. These exist because the possibilities of the technology attracted people with all kinds of interesting problems and, quite frequently, just raw curiosity. This is also the case for the iPhone, which opened up all the possibilities of mobile development, whilst preserving the relevance of existing Apple developers’ skills and offering a simple path to market.

Shared problem communities start with a very specific need; I need to get data out of a Web hosting firm that is about to shut down, or visualise water management information for northern Senegal without needing to spend $10,000 a seat, or find an alternative to Microsoft Internet Explorer. The first of these led to Archiveteam, the second to Agepabase, and the third to Mozilla. Exasperation with the telecoms vendors’ products for enterprise voice was what inspired the creation of Asterisk.

Salesforce’s Force.com is a successful example of a problem-specific developer platform; you’re using Salesforce and you have a problem that involves CRM, so off you go to Force.com. You’re trying to solve your problem using voice? Perhaps you might want to try Ribbit, which is an opportunity-based developer platform.

And this makes sense; after all, solving the problem is where the economic value emerges, and it’s the application of broad general purpose commodity technologies to very specific business problems that we want all those developers to bring to the show to extend the value.

Litmus: Neither fish nor fowl…?

But there’s a disconnect here relating back to Litmus; communities that form around the possibilities of a particular technology tend to be generalist, global, and attached to the technology rather to any particular operator or even vendor. Communities that form around a problem are more particular. Neither of these fits Litmus, although you could perhaps say that it’s about the possibilities of telco APIs in general.

It’s all rather reminiscent of J. P. Rangaswami’s notion that the more general-purpose the technology, the more appropriate it is for open source because it can scale better; a technology-motivated community needs breadth and scale.

So, while there is often value in keeping a test tightly managed as a centre of innovation and learning as O2 UK appear to have done, perhaps Telefonica / O2 will eventually be better off looking at enterprise problems and being less centred on the O2 brand name, or else broadening the possible addressable market by rolling Litmus out to the whole of Telefonica, if possible, including the Latin American markets as well. Brazil has one of the world’s most vigorous hacker communities – they invented Commwarrior, the first mobile worm, after all. Surely there’s innovation to be had there? And it’s absolutely vital to the success of the whole project that it finds a sufficiently large user elite of its own to support the developer core group.

At the moment, though, at least going by the content of a recent call the Mobile Entertainment Forum held, O2 seems to be mostly interested in using the Litmus APIs for content, rather than applications. For example, the key use case for their roaming status API is that content providers can avoid serving content licenced in one territory into another. This is fair enough as content applications may be part of the solution for Litmus, but we’re slightly concerned that they may be stepping into the vortex of content obsession, like so many other people in the industry.

Our view is that, as much as we like many elements of Litmus, in its current form and scale Litmus may well show some useful test results but probably won’t develop into a successful platform business. Building a much bigger user base should therefore be Telefonica / O2’s top priority for Litmus – even if the developer community is the key target audience. No amount of good new apps can deliver this in its current form and broader success will take good implementation of the kind of radical business model innovation we’ve outlined here.

One option would be to look in the other direction. The existing version of Litmus is targeted on consumers; what about enterprises, or small businesses/power users? This would require a different approach to signing up both developers and customers – in fact, it would be rather more app-store or app-market-like than the current “X-Factor for developers” model, although perhaps there could be a version in which the customers’ problems competed for solutions from the developers. In fact, according to Jose Valles Nunez, Telefonica is indeed considering a “business class Litmus” in the foreseeable future.

A further question is one of credibility. Attracting developers to use a platform requires their confidence that it really will be promoted and that it will stick around – no-one wants to put effort into something that might disappear in a few months’ time. Several hosted Web application environments have already done this. At BT, spending money on Ribbit was intended to act as what biologists call a costly signal – a signal that is credible precisely because it requires a real investment. Perhaps the first few “picks” for the mainline O2 Active lineup, or the first Telefonica Ventures investment, out of Litmus will light the blue touchpaper?