A Hypothesis Test at Vungle: Harness the Data Jungle

San Francisco-based advertising company Vungle provides a platform that embeds short video advertisements in mobile apps. Every month, more than 100 million people view a Vungle ad.

In 2014, the company was faced with a decision: It had a new algorithm that would use historical information to calculate what ad to show users, but implementing it would be expensive. Leadership had to be sure that the innovation would deliver results and conversion rates would rise.

To determine whether to proceed with the new approach, Vungle performed a split test, using randomly assigned users to compare the effectiveness of its existing algorithm with the proposed change. With 30 days and millions of observed user reactions providing a wealth of data in favor of the new method, the company forged ahead.

In the past, a firm considering a change might have relied on carefully chosen focus groups and market studies, investing considerable resources into the process. These days, technology offers companies direct access to customers and their preferences. The ease of data collection is shaping innovation, as hypotheses can be tested in more efficient and thorough ways than ever before.

The preceding is adapted from Darden Professor Yael Grushka-Cockayne’s article “How Vungle Lets Its Customers Innovate,” which appeared in the 20 September 2015 issue of The Washington Post as part of the Darden School of Business/Washington Post “Case in Point” series.

About the Faculty

Yael Grushka-Cockayne

Grushka-Cockayne’s research and teaching activities focus on decision analysis, forecasting, project management and behavioral decision-making.

As an expert in the area of project management, she has served as a consultant to international firms in the aerospace and transportation industries. She is the secretary/treasurer of INFORMS Decision Analysis Society, a U.Va. Excellence... Learn More