In 2015, for the first time in history, Americans spent more time on mobile devices than on desktops and laptops. But the current mobile analytics infrastructure is still two steps behind the consumer. In 2016, I predict we’ll see a number of significant changes in the space.
The transition to mobile has shifted the revenue streams away from advertising and towards directly monetizing their end users. A successful mobile business needs to understand how their users are engaging with their product, which means not only understanding what they’re doing, but also why users take a specific action. That’s where analytics comes in.
So what does all this mean for analytics in 2016?
Behavioral cohorting will become the new predictive analytics
Predictive analytics can be powerful when done right. For example, companies like Netflix and Kayak use it to suggest movies and predict cheap airfare. However, implementing predictive analytics relies on machine learning techniques – not all of which are able to produce consistent results. This puts an artificial barrier between the data and the rest of the company. As a result, insights are available only through the analyses by the company’s data science team.
In 2016, a new type of analysis will emerge: behavioral cohorting, which uses historical data to group users by actions they perform. Unlike traditional acquisition data cohorting, which only looks at the retention of users based on when they sign up, behavioral cohorting identifies key user actions that may lead to long-term retention, engagement and monetization.
Traditional cohorts don’t provide actionable information: knowing that users who joined the week of June 3 have poor retention doesn’t provide any insight into what the problem is or how to fix it. On the other hand, using behavioral cohorting to discover that users who never perform a specific action are more likely to stop using the app may indicate that this behavior is important for retaining a user long term.
Behavioral cohorts help teams understand when and how users experience their product’s core value and identify which areas need improvement. Traditionally, behavioral cohort analysis is just as time-consuming and complex to execute as predictive analytics, requiring data scientists and SQL queries. But in 2016, I expect that advances in analytics infrastructures and user interfaces will make behavioral cohort analysis more accessible and widely adopted.
Complex queries will be available through simple user interfaces
Understanding user behavior used to be difficult and expensive, requiring engineers to run complex queries for behavioral cohorting on terabytes of data held in memory. The time lapse from asking a question about user behavior to getting meaningful insights to implementing a product update would take weeks or months, resulting in slower product iteration and, ultimately, slower growth.
2016 will also be the year when complex SQL queries are executed through user interfaces. Organizations need a UI that allows anyone, regardless of their knowledge of SQL, to answer their own questions in real time, by creating personalized cohorts with parameters as complex as needed.
Data visualization will be more interactive
Visualization is critical in understanding key user trends, but existing tools often don’t provide enough flexibility to see specific metrics such as how users are behaving in an application. Data visualization products like Tableau and Qlikview have moved the industry away from traditional graphs and dashboards constructed using R or Microsoft Excel. However, the traditional analytics landscape is still largely comprised of static aggregations of raw data.
In 2016, visualizations will become more interactive, allowing organizations to manipulate more than just date ranges on the x-axis of a graph. Instead of only seeing overall trends, tools like Amplitude’s Microscope feature will give analytics users the ability to click on any data point and dive into the individual users and behaviors that are hidden within the big picture.
As consumers continue to shy away from Web applications and traditional methods of monetization like advertising, understanding user behavior within an application will become critical for the success of any business. 2016 will introduce more self-service, decentralized analytics platforms that will make complex behavioral insights accessible to non-technical end users at companies of any size and budget. This will enable everyone to leverage data to create the best mobile experiences possible and drive maximal growth.
Spenser Skates is CEO and co-founder of Amplitude. He founded Amplitude in 2012 to revolutionize the way companies interact with their mobile users. He and co-founder Curtis Liu completely rebuilt the mobile user analytics stack, creating a behavioral analytics engine that allows companies to understand what their users are doing in their applications. Prior to Amplitude, Spenser founded Y Combinator startup Sonalight, a voice recognition app. Earlier, he worked as an algorithmic trader for DRW Trading Group.