You can "fake" mobile support using Twilio. It will not have an experience as good as a native smartphone application, but the cost is lower by (plural) orders of magnitude, and the reach is far greater.

I'm writing Appointment Reminder, which features a scheduling interface (in a web app). I have heard from customers that it would be really, really handy if they could access their schedules while away from a computer. One of them specifically asked for a Blackberry app.

I could write a Blackberry app. And an iPhone app. And twelve Android apps. And... no, wait, this is insane. I have enough on my plate just getting the underlying service and web interface to work.

Instead, I am writing something responsive to the underlying need ("I need to check my schedule when not at a computer") with Twilio. Call the number you programmed into your phone, hit 1, bam, here is your schedule for today. That gives me 100% coverage on all handsets, including feature phones, for approximately an afternoon of work.

Now, if that feature blows the doors off, it might be worth surveying my customers about what phone they have and implementing e.g. an iPhone, BlackBerry, etc app as resources permit. If nobody uses the feature at all, or if I can't get the marketing for Appointment Reminder to the point where I have any customers to worry about not having an iPhone app, then I didn't spend several tens of thousands of dollars in vain to learn that.