I have been using Pandora’s online stream music service off and on for several years. What got me more interested lately was it being one of the many services on my Roku video streaming box, which my wife and I use mostly for watching movies from Netflix’s "watch instantly" queue.

As I investigated the service more, I came to understand exactly the challenge of what it takes to be truly multi-platform in the current era. It isn’t just about having both Web and mobile phone versions of your service, but how you have to go deep into a lot of different devices to appeal to your customers. The cool thing about Pandora isn’t that you can create your own custom radio station that will try to find music based on a particular artist or genre. But that once you set up your account on one platform, you can access it in your car, in your home, and on the road in between.

All with the same collection of stations and music.  As you spend more time with the service, it tries to figure out your likes and dislikes. Let’s look at all the various places you can get your Pandora fix as an example of how hard it is to become this ubiquitous. First is the Web browser: you have to work in a bunch of them properly, so there is the usual testing in IE, Firefox, Chrome, Opera and Safari. Add Mac, Windows and Linux versions of each browser, and that’s 15 regression tests right off the bat. But we have just gotten started. Add in the newer brower versions, like IE8, the fact that Linux isn’t a single OS, and 64 bit Windows. Then stir in support for both Flash and HTML v5, and you can easily get more than 200 different environments if you want to support a wider base. Pandora, by the way, doesn’t officially support much beyond Flash on Firefox, IE, and Safari on Mac and Windows.

Multiplatform Compatibility Doesn’t Stop There

Then we have separate apps for each of the five mobile phone platforms (Blackberry, iPhone, Android, Palm Pre, and Windows Mobile) and four cellular providers because their phones work differently on each network. Never mind that each phone’s ecosystem has different rules on how an app can get posted for download and get itself updated. There are at least twenty different tests there. The phone apps have to be designed to work with the limited screen real estate available on each phone, and yet still connect to your account in a way that you can recognize without a lot of user training. Some of the phones have different screen and control button configurations, so just supporting the Blackberry line, for example, isn’t so simple. You also need to get the development environment for the phone (typically these run on PCs with simulators that show you what your phone user will end us seeing) and probably a bunch of phones to test out too.

But wait, there is more. How about Facebook, My Space, and other social networks? Don’t you want to integrate with them and leverage them to make your app viral? More code to write, more interfaces to learn, more tests to run to make sure you new versions don’t break these links.

Then there is support for the home-based entertainment systems. While each of these have some embedded Web browser in them (like the Roku or the Samsung BluRay DVD players), you still have to test to make sure that the pages load properly and the music keeps on playing and your fancy navigation controls operate as intended. There are more than a dozen different devices, including the Ford Sync in-car service that will be available later this year, to test out. The trouble here is that these devices typically have older and less capable browsers that don’t get updated, unlike the PC world where users are trying out new versions.

As you can see, it is easy to lose count of how many different platforms you want your app to run on. And then if you have to make choices and limit yourself, how do you do the triage? Do you drop Andoid in favor of Roku? Bring up the new Ford Sync API and leave the Pre to wither away? The user populations of each of these communities is constantly changing, as sales wax and wane.

It is enough to make many of us long for the simple days of the 1990s, when we just had to worry about Mac vs. Windows support. I got the idea to look at Pandora from an article in today’s NY Times. And while the service can wreck havoc on corporate networks (lots of folks start the audio stream and then walk away from their PCs), I think they are doing exactly the right kind of things when it comes to managing their multi-platform strategy.