The hidden costs of developing mobile apps in the cloud
You've developed a hit mobile app. Congratulations! Now get your wallet out.
It's one of the ironies of the developer community that being successful can come at a price--a steep one if they have opted to use a cloud-based service as part of the process of creating and running an app. That's because, unlike the days of on-premise software that developers might pay for once and then use however much they wish, cloud-based tool sets offer the ability to scale from hundreds to thousands (and perhaps even millions) of concurrent users more easily. However, in some cases the more users a developer attracts, the higher the costs for the cloud provider, who must keep running the infrastructure to support a larger user base. The cloud providers, in turn, pass those costs back onto the developer.
To prevent things from getting wildly out of control, companies tend to package cloud-based offerings into tiers of usage. This means developers might need to estimate--some would say guestimate--how many users they believe their app will attract once it's launched. If they're wrong, they could end up paying a lot more than they bargained for.
appMobi offers cloud-based tools for developers, including for Web-based and hybrid-HTML5 apps.
Earlier this month San Francisco-based appMobi tried to set itself apart from the competition by offering to make its cloud-based tools – which help developers make apps in Java and HTML5 – free until an app becomes profitable. For appMobi, that means any app with less than 10,000 users per month. After that threshold, developers will have to pay 10 cents per user.
How to evaluate your pricing options
Roy Smith, AppMobi's vice president of marketing, said the traditional approach to cloud-based pricing came from the early days of Web hosting, where site owners had to forecast a certain amount of traffic to their hosting provider. That model doesn't really work in a world of smartphones, where the definition of success is more about using a piece of software than simply visiting a website.
"A developer might go for (a package based on) 500 active users. But then what happens if Apple puts it on the Featured page in its app store? Suddenly that developer has a big problem," Smith said. "We're trying to reduce the friction for the developers to be able to stretch their capacity if an app blows up."
Others firms are also trying to ensure developers aren't penalized for creating the next Draw Something. Redwood City, Calif.-based Sencha, which provides HTML5 frameworks for both mobile and desktop developers, is primarily working in the enterprise app space, but a growing number of its users play in the consumer market as well, said Aditya Bansod, vice president of product marketing. To accommodate a wide range of developers, Sencha plans to offer its forthcoming cloud-based services (which are still in beta) in several ways. These include a free tier so developers can gauge how their app is performing, as well as a paid service.
"The consumer applications are much more like public media, if you will. There's a long tail. But where the spike is, it can be very large. We're allowing an affordable option in the long tail," Bansod said. "We don't want to be a complete burden on your own success. We want to share in your success."
Buying usage as it's consumed
Mountain View, Calif.-based Appcelerator, meanwhile, claims more than 350,000 mobile application developer customers that use its Titanium platform to create native iOS and Android apps and HTML5 mobile Web apps. Michael King, a director with Appcelerator, said the company also offers a "freemium" tier, which allows for up to approximately half a million API calls. Depending on the app, there could be up to five API calls per user (to do a check-in, for example, or to process a user name and password authentication), so free usage might top out at a couple of thousand users and 5 GB of storage. After that, cloud-based services are charged on a per-use basis, which King said ultimately makes the most sense.
"They don't have to predict the impossible. They can really buy (usage) as it is consumed," he said. "It really is viral in terms of creating a successful app. It isn't until you have at least a certain critical mass that you see massive amounts of adoption."
Besides, he added, if developers were using Amazon Web Services or setting up their own servers, they would have to predict how much capacity they needed, along with all the coding costs, licensing and virtual machine usage.
Bansod agreed, noting that there is a difference between working in a private cloud environment where the developer takes on all the costs vs. a public cloud where the operating expenses are taken on by a platform provider. But as apps become popular, the economies of scale can work both ways, he added. "As you become successful, the per-unit pricing will start decreasing," he said.
Getting to critical mass is key
Of course, for most developers having a runaway hit app that costs them more in cloud service fees would be a nice problem to have, and no provider will charge so exorbitantly that the developer will move to another platform. However, Smith suggested that developers will find the most value in those providers that can give them not only the tools to create an app, but to help them reach critical mass more quickly. That's why appMobi is offering Webinars and tutorials on YouTube to give developers more insight into how to effectively use analytics; Appcelerator is running online classes on how to develop more compelling UIs and user experiences.
Whatever the approach, the number of developers attached to a platform matter to the cloud providers--they need critical mass as much as anyone else.
"There's really an exponential curve--of the millions of apps out there, there might be only a couple of thousand that will really go through the roof," Smith said. "We think if we have more at the bottom of the funnel, we'll have a better chance of supporting whoever comes up with the next Angry Birds."