Having recently become an iPhone app developer, I've been wrestling with a problem that continues to vex mobile developers on various platforms: marketing my app.

Were Blocfall a spectacular iPhone game, marketing would be easier. But it's not. It's a modestly entertaining diversion from someone who is better at writing for people than for machines. It's no Angry Birds. Maybe next time.

The inherent difficulty of convincing someone to download an unknown app is compounded by competition. There are a lot of apps out there, over 300,000 in Apple's iTunes App Store and maybe half that in the Android Market. And many of the obvious opportunities for exposure, like submitting an app to a review site, don't always work. Too many review sites want to be paid for a review. Better to buy an ad, which is what a paid review is.

My admittedly naive hope was that I could generate interest through online advertising. But traditional text-based and display advertising turns out not to work very well, at least not for encouraging people to download, install, and use an app. Cost-per-acquisition (CPA) or cost-per-install (CPI) advertising, however, is a different story: I made money, on paper at least, in the form of unexpected free advertising, using Flurry's App Circle CPI program.

Clicks Are Worthless
The first thing I learned when I tried to advertise my app is that you can go broke paying for clicks. I'm sure there are some circumstances when you might want to run display ads. But trying to encourage people to install your mobile game is not one of them, unless perhaps you're Electronic Arts and you're happy to burn cash to sustain brand awareness.

Late last year, when Blocfall was released, I paid $100 to AdMob to run a text ad on iPhones to promote the game. The ad was shown 324,960 times, generated 2,000 impressions -- a 0.62% click-through rate -- and led to maybe two or three sales of the then $0.99 cent app. A dismal failure.

More recently, I bid $1 per click on a Facebook ad, a campaign that from 86,941 impressions generated 23 clicks -- a 0.027% click-through rate -- at an average rate of $0.58 per click. The total cost of the campaign came to $13.26 before I pulled the plug. The Facebook ad resulted in no installations of the app, so far as I can tell. It was another dismal failure (and it made me wonder about Facebook's sky-high valuation).

My experience with Flurry started off on a similar trajectory. Flurry offers developers analytics tools -- I can insert Flurry code in an app and get anonymous data about how the app is used -- and CPI advertising.

Having paid the $250 minimum to participate in its system (a bit too high if you ask me), I set up a CPI campaign for Blocfall. I bid $1.30 per install. In other words, every time an iPhone user sees a Flurry ad for my app, clicks through and installs it, Flurry is supposed to charge me the price that I bid for the app installation. I get a user, Flurry and the publisher that showed the ad get to share the fee.

I raised the price of Blocfall to $1.99 in the iTunes App Store in order not lose money: With Apple paying me $1.40 for the sale of my $1.99 app -- for which I'd be paying Flurry $1.30 -- I would net $0.10 per app. You have to start somewhere, right?

But it didn't work. With 26,510 impressions generating 575 visits to a learn-more page and 64 visits to the iTunes App Store, the campaign generated only 1 install. Thankfully, that experiment didn't cost much, with only one installation being billed.