Splitforce Updates Toolsuite for Mobile A/B Testing
Behavioral testing of mobile applications is becoming more and more important for a huge number of companies. Even if most companies are not yet going real "mobile first", mobile apps drive relevant parts of their businesses. Splitforce launched a tool suite to optimize mobile applications by A/B-testing in 2013. Now, Splitforce launched an updated version of its tool suite. Besides an improved user interface, the new versions offers new functionalities like user-targeting, tests based on behavioral data or auto-optimization.
A/B-testing is commonly used to optimize web sites. Developers prepare variations of the site which differ in minor aspects like color schemes or button labels. These variations are presented to users in a predefined ratio to identify the more successful variation regarding conversion rates, timing or quantity goals. The most successful variation is then presented to all users of the site.
Whereas this is an easy task for conventional web sites since server-side code can be updated anytime and changes are reflected immediately in the users' browser, native mobile apps are distributed via app stores and cannot be changed on-the-fly. Splitforce lets registered developers define experiments and goals and offers native SDKs for iOS, Android and Unity to embed variations in their apps. At runtime, those SDKs connect to Splitforce' web services to find out which variation to show to the user or to notify Splitforce of a fulfilled goal. Eventually, developers can activate to most successful variation for all users without re-submitting the app to the app-store.
Now, Splitforce introduced various new features to improve the process of mobile app optimization:
- User targeting: Instead of randomly choosing variations at a given ratio, variations can be chosen based on certain conditions. Conditions can be divided into "predefined conditions" like screen size, language settings or operating system versions and "custom targeting conditions" which refer to any information present in the customers' user database.
- Tests based on behavioral data: Variations can be selected based on data from previous usages of the applications like session duration or time of last app launch.
- Auto-optimization: With auto-optimization, more successful variations will automatically be shown more often.As soon as auto-optimization is activated, a learning algorithm observes the variations' performance. The higher the observed performance, the more likely a variation will be picked for the next user. Eventually, suboptimal variations will automatically be pruned.
InfoQ also got in touch with Zac Aghion, CEO & co-founder of Splitforce to comment on the new feature of auto-optimization:
From a high level, the way that auto-optimization works is that we start showing changes or variations that are performing better more often automatically. For a product manager, the way that you would be able to use this, is to load up different variations that you believe would have an impact on the performance of your product and then basically just hit "Go!" on auto-optimization. And then changes that are performing better are shown to a greater percentage of your users automatically without you having to touch any dials.
Spliforce' service plans range from $20 per month for basic functionality to $500 per month for five million data points and unlimited applications and users. Companies can also apply for a custom enterprise plan which also offers more support and services than the pre-defined plans.