Dev Watch

Blog archive

Mobile Developers Digging Deeper with App Analytics Tools

Hmm, KitKat 4.4 mobile apps are crashing at a rate of 2.45 percent, slightly better than the newer Lollipop 5.0 OS, currently at 2.23 percent. But Lollipop has been volatile lately. Let's check tomorrow's delta and decide whether to pull a coder off KitKat to work on Lollipop fixes.

Welcome to the new world of mobile app analytics, where developers can track how their apps are performing, how the different OSes are performing, how networks are performing and just about how anything else that can possibly be measured is performing -- with a few clicks.

Or one click.

For example, click here and you can see those exact aforementioned stats and a bunch more, courtesy of a real-time dashboard produced by Crittercism. The mobile application performance management (APM) vendor has been highlighting its service lately, trying to make PR waves in advance of the upcoming Mobile World Congress 2015 conference in Barcelona. The company earlier this month signed a strategic partnership with digital optimization company Keynote and last month hired a new product management exec, putting to use some of the $30 million in financing it garnered last year.

"The real-time dashboard leverages Crittercism's data -- gathered from processing over 30,000 apps across 1 billion users -- to provide both developers and businesses up-to-date insight into OS adoption and performance, as well as average Wi-Fi and carrier latency worldwide," the company said.

The San Francisco firm, founded in 2010, is jostling for position in an increasingly crowded market.

For example, Twitter, currently in a traveling road show to attract coders to its mobile app platform, on Tuesday reported its own app analysis tool, Answers, is catching on with developers. "Answers now processes over 5 billion sessions per day -- over 1 million analytics events per second -- and is used by some of the largest apps in the world," it said.

Some of the more popular services in the app analytics space include Flurry, Apsalar and Localytics, judging from a Quora discussion that includes input from actual developers on the front lines.

Flurry, acquired by Yahoo last year, just a few weeks ago launched a new developer documentation site. Flurry said it "provides more than 170,000 developers the business data they need to understand their audience, usage and performance." It said that data is collected from 150 billion app sessions per month. (Update: Yahoo at its inaugural developer conference today introduced the new Yahoo Mobile Developer Suite, which it said will help developers "measure, monetize, advertise and improve your apps with Yahoo tools," specifically, Flurry.)

Among the dizzying array of competitors are a variety of pricing plans, service levels, capabilities and specific niches. Apsalar, for example, is one of the many vendors focusing on marketing rather than crashes and other metrics that might interest front-line developers.

Localytics, a Boston company founded in 2008, offers an analytics and marketing platform that "currently supports more than 5,000 customers and reaches 25,000 apps, 1.5 billion devices and 50 billion data points monthly."

And, of course, Google Analytics offers a mobile service of its own, with the requisite crash and exception reporting. "No one likes a buggy app," the company said. "With built-in crash and exception reporting, Google Analytics lets you triage and prioritize the issues that are affecting your users, helping you to build a better experience."

To help developers investigate the many analytics options available, Apptamin provided a tool roundup, where it discusses many more companies, including Mixpanel, Countly, Appsee and others.

Whatever service you use, the experience is bound to provide data that improves on the old way of tediously sniffing out bugs and performance problems by hand, helping out with tasks such as code analysis.

Just ask Facebook. The social media giant usually develops its own tools for in-house use, but clearly exhibited the benefits of systematic testing and bug-sniffing in its detailed account of how it tracked and fixed a nasty bug that was corrupting files on iOS.

"One of our top crashes on iOS was manifesting itself in Apple's Core Data system, an object-relational mapper into the underlying database (SQLite)," the company said. "We were receiving these crashes into our crash report analyzer, but it took months to figure out the right angle to approach the problem."

Using some in-house tools called Hipal and Scuba, and its own A/B testing scheme, Facebook finally solved the confounding problem.

Finally, the company said, the fix "cut the Facebook iOS app crash rate in half and put a longstanding problem to rest. It turns out that abandoning manual code analysis was a good strategy."

Posted by David Ramel on February 19, 2015