Google doesn't like to guess. The web giant famously tests the effectiveness of everything, from the placement of ad blocks on a search results page to the color of its logo.
To do this, the company uses something similar to focus group testing: Different users will be shown different versions of a website, and the company tracks which versions perform better. It's called A/B testing, and it's now being used by everyone from tech startups to political campaigns.
Momchil Kyurkchiev and Andrew First did hundreds of these tests for Google. Now, their company Leanplum, which launches today after a year in beta, wants to help bring the same kind of thing to the rest of world, offering A/B testing tools for developers building mobile apps.
The pair are following in the footsteps of fellow Google alum Dan Siroker, who founded Optimizely, a company that sells A/B testing tools for the web. In fact, there are now many A/B testing tools for the web, and plenty of analytics tools for mobile that give insight into how customers are interacting with applications. But Leanplum is focused squarely on mobile A/B testing.
"On the web," says Kyurchiev, "You can just throw a switch and start sending different data to some users." But today's developers want to build native mobile apps that must be installed on phones. That means either submitting each and every testing change to the app store one at a time, or including several variations of an app within the same file, which leads to messy code.
To solve this, Leanplum provides much of the application's logic via its remote servers, delivering variations of a particular application through its application programming interface, or API. The service is available for iOS, Android, and JavaScript applications. "We highlight the statistically significant results in green," Kyurkchiev says. "You don't need a data analyst to interpret the results for you."
But he does caution users about taking A/B testing too far. Some users, he explains, would do an A/B rather than just asking a customer if they wanted something, and that just doesn't make sense. Sometimes, low tech is better.
