“What are we going to do with all that data?” That was the question I asked of my co-founders at the end of another long day spent chipping away at what was to become the platform we built our company on: SAMI, our Smart Application Management Infrastructure. We were discussing the importance of analytics, and how we wanted to track not only whether or not users were logging in, which was arguably a standard in our space, but also to go much further. We were going to track every user interaction with the app, including swipes, taps, and time spent on features. In essence, every action that the user took in the apps would be recorded and compiled within our data infrastructure. What we were going to do with all that data was the question at hand. The answer was a good one: “We don’t know, but we’ll figure it out.”
Data Drives Strategic Decisions
And figure it out, we did. Not only did obvious uses present themselves, such as basic session information which we could use to determine active users, and data entered such as amounts which could be used to compile money movement statistics, but broader topics were presented as well, such as common feature usage, feature completion ratios, common usage patterns. As we began to compile large amounts of usage information, we were able to use that information to drive product decisions, such as creating new and easier ways for users to login based on login failure rates that we were tracking. Additionally, since we were making that information available to the financial institutions that we serviced, they were able to use that in order to make broader strategic decisions, such as what features to continue to invest in based on usage. Since the usage data was granular enough that it could be taken all the way down to the individual user level, we were also able to start making proactive decisions on behalf of the users, such as bringing commonly used features to the forefront after login.
Data Fuels Security
An interesting angle on our big data story is, of coarse, security. How could we use that data to further secure our user’s sessions, as well as our customer’s services to their customers? What sorts of decisions could we make using that data? Decisions around fraud, device status, and user authentication? We began to tackle all of these questions and more and quickly discovered that, as with usage analytics, our data was instrumental in helping us answer fundamental security questions.
One of the areas that was fascinating to me was what were able to determine about users given their behavior in the applications. For example, for a given user, doing remote check deposits, over time a tremendous amount of usage data is compiled, which in aggregate helps us determine a basic, common usage pattern for a particular user. When an anomalous event takes place, we can make a determination about the situation given all that we already know about the user’s behavior in previous sessions and determine whether or not this particular anomaly is likely to be fraudulent. Additionally, since we are able to track usage at an even more granular level, given the device’s sensors, we’re able to begin to make even more specific determinations of who is in fact using the device, based on a series of sensor readings and analysis of the data. When I type in my password, over time, a determination can be made about how I type in my password, given the cadence of text keys entered, the pressure used, and various other factors. Once I’ve been identified, the same algorithmic approach, using the same data, can be used to verify that another user may be entering in a password, even if they are entering in the same password, because they are not able to perfectly match how I enter my password.
Data Enables More Intelligent Applications
What innovations like this begin to open up is a new world of authentication. Given the rise of biometric solutions as the next step in user authentication, a slight variance of biometric, in this behavioral biometric, holds a tremendous amount of promise. In addition to new methods, such as TouchID, voice and facial recognition, behavioral techniques are a new promising way of being able to layer on yet another set of data to help determine who the user is. There are many uses for this technology that can even span outside of specific security solutions. For example, applications that respond to how users are interacting by making some screen elements more prominent, making predictions about data entry and pre-filling fields based on past entry, etc. Having the data in the first place was only the beginning. As we continue to discover, data is powering more and more of the decisions we make, and is enabling our applications to continue to get smarter.