Open platforms -- APIs are not enough

Five years ago, everything in mobile was about the flood of innovative applications we could expect from 3G. All eyes were on the services space. The reality was that 3G data was more about driving handsets sales (feature ticklist), and the high margins were in the "dumb pipe" business of roaming mobile enterprise users (datacards and Blackberries).

The space between these two extremes is the platform business that a few telcos are only now just beginning to embark on creating. Keep yourself in the value chain by being a supplier of identity, payment, content delivery, personalisation, advertising, and adaptable wholesale connectivity capabilities. Enable a partner ecosystem which addresses the myriad niches a one-size-fits-all product strategy can't reach.

Over the weekend I was installing some extensions for my email client, and I thought it would make a good case study in failing to drive home strategic advantage of a platform strategy because of a poor user experience and execution. In this case, Mozilla (who make the Firefox browser and Thunderbird email client) have screwed it up on an epic scale. Learn from their mistakes.

This is important. The real threat to operator cash cow services from Internet players does not come from VoIP arbitrage (e.g. Skype), or even the stagnating portal players (Yahoo!, MSN, et al) and their hybrid IM/VoIP clients. It's from social networking and self-expression companies like Facebook, with deep relationships with their customers (far deeper than telcos or traditional nemeses like Nokia). They're learning to open up their platforms. As one VC puts it, choosing between supporting a platform-centric partner-friendly business model, and one that isn't (like, say, a telco) "isn't a decision, it's an IQ test". You'd have to be stupid not to choose the former.

Ultimately every social networking or commerce hub will seize the user interface (via the buddy list or directory) and control who gets the revenue from communications. Today, the fight is on in every segment to become the hub for relationships -- kids, teens, twentysomethings, B2B, consumer to SME and consumer to enterprise. Some of these may become "the next Microsoft" as impossible-to-bypass pinch points in the communications value chain.

From Cluetrain to Smart Mobs, to Communities Dominate Brands -- the runes all spell the same thing: open up, engage others in co-creation of value, share the rewards.

Anyhow, enough of strategy. Back to execution.

The browser wars between Microsoft and the open source community (as offspring of Netscape and Google) have gone on for a long time now. Mozilla's Firefox browser has recently been picking up a significant market share from Internet Explorer, just crossing 25%. One of the key reasons it that the core browser was kept relatively lightweight, and then you could add on the extra capabilities that you wanted -- rather than shipping a thousand buggy, insecure bloated features that (almost) nobody wants. They've then taken this model to other products, most notably their email client, Thunderbird.

So having heard that there's a new calendar feature, I thought I'd check it out. Now, remember that there are only three things that drive people to use Firefox and Thunderbird:

  • Irrational dislike of Microsoft. Small but vocal minority.
  • Desire for standard browser features like tabbed browsing and faster rendering, which Microsoft have largely matched in IE7 anyway. (Thunderbird does IMAP email access well, whereas Outlook is a disaster when not coupled to Exchange.)
  • Some extension or other capability you can't otherwise get. The only place to go is Firefox or Thunderbird.

Extensions tightly lock users and developers into the Mozilla ecosystem, as well as driving the user needs and product roadmap (as top extension features get folded into the core product). Anything that stands between a user and an extension installation is a strategic disaster.

So, let's say I've heard that there's a new calendar extension for Thunderbird that I might like to try out. I go to Google and search for "thunderbird extensions".

First problem. Mozilla has created two different places on their own web site. Which do I choose? Lose 25% of your installs at this point. Click on the second one -- I'm looking for extensions, not add-ons, after all.

When we get there, there's another choice of two places to go, with no indication of which one to follow. When we get through to the extensions catalogue, there are some categories, but none of them obviously correspond to calendars. The only search box is a generic one for the whole site, not just extensions. And there are far too many extensions in each category:

Do they seriously expect you to page through all these looking for what you want? Furthermore, the calendar application from Mozilla themselves is cryptically called "Lightening" -- not "Mozilla Calendar", or even "Mozilla Lightening".

Now we get to the really horrible bit. Glance down at the next image for just a moment, and what's your first reaction?

Same as me, I guess? Click on "install now". Nice try...

The instructions above tell the story. Don't click on the "install now" button! Nope, we've not worked out how to package up extensions that aren't for the browser, but rather for the email client. "Install now" tries to install the email client extension into the browser, which is rejected. This is embarrassingly poor. (I've still never got the British English dictionary working in the email client as well as the browser).

This isn't a localised problem. The #1 strategic lever Firefox has in the mass market over Internet Explorer is these extensions. Surf to an extension outside the formal catalogue, and what do you get?
Click to view.

A pointless message telling you that we're not going to ask you if you want to install this. Why not just damned ask "Click here to install."? Even worse, the only option you get when you click the box is this:

Permanently approve all future requests to install extensions from this site! Madness.

This isn't the only set of things that are wrong. For example, you can't navigate through to extension preferences from the preferences menu -- you have to be telepathic and know that the options for extensions are under the add-ons menu.

In summary, you need to get several things right if you're building such a partner platform:

  • Consistent branding of platform-enabled services
  • Clear and consistent terminology
  • Search-engine optimised portal or partner sites
  • Easy navigation and search of partner services
  • Constrained choice, or at least some kind of recommendation to users.
  • Simple user experience for provisioning.
  • Certification program that separates "own brand", "approved" and "other" categories of partners (a-la i-mode).
  • Proper integration into the self-service and management capabilities of the base platform.
  • Security model and certificates for all parter extensions that don't just let them romp all over the user's device, but declare what access you're granting them.

There are some things that Mozilla get right. For example, the system for managing upgrades of extensions is superb. But five years from now, Mozilla's application suite will have gone the way of Netscape -- probably trampled on by packaged custom browsers aligned to each user's favourite social networking hub.