Managing Application Failure in the Mobile Marketplace

The rise of the mobile app economy has brought along a number of challenging hurdles for developers. No matter how much internal testing is done, until an app is released out into the wild, there’s no telling how it will truly perform under real world circumstances.

I caught up with Andrew Levy, CEO of Crittercism, a provider of end-user Application Performance Management (APM) software, to discuss the top reasons for application failure and the costs associated with unexpected mobile app behavior.

Levy identified five major failure potentials:

1. Device Specific Problems – Each mobile device is different, and as a result, applications may perform differently based solely upon the specific model on which they are run.  
2. Operating System Specific Problems – Particular operating systems may not be compatible with the application, or may have bugs that can cause issues with certain programming libraries.
3. The Application Code Itself – The application may simply be coded improperly or inefficiently.
4. Unexpected User Behavior – There’s no telling how a user may attempt to use an application or how many users will attempt to use an application at a single time. Unexpected loads and use cases can drastically affect an applications performance.
5. Network Issues – An application can run into carrier specific problems, problems with cloud services, or failing APIs. A lot of apps are actually thin clients that depend on these services to operate properly, and without them, the application is useless.

Determining precisely which of these five areas are to blame for a malfunctioning app is nearly impossible without the help of APM software, and the costs of failure can be staggering. According to Levy, the typical application experiences around 2 percent average downtime. For an e-commerce business like eBay, which processes more than $20 billion in transactional volume a year, 2 percent downtime can equate to $400 million in lost potential commerce. Of course, a poorly functioning app will also led to poor app store reviews, further diminishing a business’s return on investment (ROI). Obviously, companies can’t afford to miss out on these types of opportunities.

Customer facing apps are not alone in this struggle, however. Enterprise mobile apps are increasingly being used to scan inventory in the warehouse, as point of sale devices, to track truckers, and to service customers at their homes, among other things. These apps can significantly affect both the top and bottom line, so performance is paramount.

“That’s the way the world is moving,” Levy explained. “MDM providers have filled the security checkbox for a CIO to actually deploy business critical functions onto smartphones and tablets, but now they have to deal with the operational complexity, and that’s where Crittercism comes in.”

The job of companies like Crittercism is to help operations and engineering teams triage these application issues. APM software sums up real-time diagnostic data about performance issues, including app launch data, crashes, errors, data pulls and network calls to different cloud services and APIs, along with the associated response codes, latency and request volumes.

“Once they understand what the total user impact is, they are given the data needed to actually go back and figure out exactly where in the code the problem came from, what user behavior may have led to that problem occurring, what device configuration caused the problem, and so forth. Only after determining the nature of the failure can adjustments can be made,” said Levy. “These decisions are extremely hard to make without actually having the data to back up how you need to deliver those experiences.”

For example, a major U.S. retailer figured out, using Crittercism software, that during a thirty-minute window at around 8:00 p.m. Pacific Time, its application was completely failing to allow users to check out, right when customers would be typically sitting on their couch with their tablets or smartphones to make purchases. Coincidentally, this equated to about 2-percent downtime, and again, a huge amount of dollars being lost. It turned out that a server process was scheduled to run at midnight east coast time, thus preventing west coast customers from completing their purchases.

Location can be a huge issue, too. Many application issues can arise simply by switching between Wi-Fi, Edge, 3G, and LTE networks. Moreover, on a geographic scale, many regions may not have the required bandwidth needed to support high fidelity streaming applications. In such scenarios, companies may need to develop a lower fidelity experience for these customers. 

Without application performance management solutions, however, these facts remain a mystery, and that’s the last thing a business wants when it comes to a business critical app. As the mobile app landscape evolves, APM solutions will undoubtedly continue to play a crucial role.



Image via Shutterstock