Interactive Behavioral Analytics for Winning in the Digital Economy

If you have read my recent post, you will know that I am interested in product management and data science. These 2 fields combined create a data-driven product development process, which is a powerful approach in a lot of SaaS companies nowadays. Recently, I went to a product meetup in Imgur headquarter in San Francisco, and had a chance to mingle with other product people as well as learn about the trendy product practices. There were 3 talks presented, all centralized on the topic of the meetup “Interactive Behavioral Analytics for Winning in the Digital Economy.” The 3 speakers are Ajay Arora (VP of Product at Imgur), Jackson Wang (Analytics Lead at Tilt), and Lior Abraham (Cofounder of Interana). The common theme of the 3 talks is how to use behavioral analytics to get an in-depth understanding of exactly what users want, need, and don’t like to develop the right strategies to attract users, increase engagement, and reduce churn. Below are my notes on their talks, which I deem to be extremely great insights for any software product people out there.

Imgur is one of the highest trafficked sites on the Internet and ground-zero for the most engaging and viral content on the Internet. As VP of Product, Ajay leads the product, design and data teams, with joint meetings and product proposals reviews. Ajay provides 3 lessons he has learned since he joined Imgur:

Lesson 1 — Trust your instincts, they are probably wrong, so verify them with data.

In particular, Ajay gives an example of Imgur user retention problem. Instinctively, he thinks that when a user attempts to register with Imgur mobile app, the user will more likely to stick with it longer. However, the data show that users are actually more likely to quit the app right after they visit the app — which can be hard to explain.

Lesson 2 — Data insights point to a problem, user research solves it.

User research comes to the rescue at this point. The Imgur team spends time talking with its users, and discovers the intuition behind the user retention problem. The user struggles with the onboarding process; in specific, he can’t create an account because all of the familiar usernames have been taken. For that reason, he quits.

Lesson 3 — Some of the biggest roadblocks can be solved with really simple techniques.

The simple solution that Ajay and his team came up with is to apply an auto-suggestion algorithm that suggests usernames based on email account. Another minor detail is to input a line on the screen saying “Don’t worry! You can always change it later.” This in fact keeps the users on Imgur app longer and they are much more likely to use it more frequently.

These 3 lessons show how Ajay and the Imgur team have been able to utilize data insights to track user behavior and come up with a solution that meet user needs.

Jackson Wang of Tilt is up next. At Tilt, he is in charge of measuring and helping to improve every part of the user experience on the website. Previously, he worked in the analytics team at LinkedIn. If you don’t know, Tilt is a crowdsourcing product that allows for groups and communities to collect, fundraise, or pool money online. Or in other words, it is the easiest way to collect money from a group. Tilt’s target demographic is mainly college campuses.

Back to the point about behavioral analytics, Tilt’s analytics team utilizes a super interesting model called “user state machine”, which depicts different phases Tilt’s users can go through and communicate: Phase 0 — New Users, Phase 1 — Organizers, Phase 2 — Contributors, Phase 3 — Visitors, and Phase 4 — Inactive Users. The main goal is to increase the number of organizers, because they drive contributors, visitors, and even inactive users. With this acknowledgement, Tilt’s strategy is to focus on organizer’s retention rate, which has been vital through the course of the company growth. Jackson concludes that the 2 main benefits of data analytics are:

· Decrease the cost of asking questions.

· Decrease the cost of experimentation.

Finally, Lior gives a presentation called “Software Bottleneck.” Lior is a former Facebook executive. He was instrumental in scaling Facebook’s analytics and growth initiatives. Most prominently, he invented Scuba — a visual and interactive analytics solution that allowed Facebook to develop innovative strategies around growth that helped them get over billion users they have today. The main theme of his talk is Software is the Bottle-neck to Insight, meaning that software programs hinder the process of coming up with useful insights.

Essentially, the state of data software 10 years ago was not good. A data scientist have to do a lot of manual metrics and pre-calculation — things like custom rollups, tracking grow with an exponent… So Lior’s goal while developing Scuba is make it easy to write and read data. In particular, Scuba has analytics-mode only, ad-hoc exploration, a raw scan, visual display and agility. To quote Lior directly, “Speed made things interactive. Interactive made data accessible.” The end result is that people began to use data differently:

· People becomes closer to data

· Teams work more cohesively

· Even non-technical people start seeing more pattern

It is definitely important for companies (especially startups) to recognize the needs of exploiting behavioral analytics if they want to get leverage in the digital economy. Powerful tools like SQL, Facebook Analytics, Hadoop, Interana.. are great solutions for engineering/product teams to handle the massive amount of user data and come up with innovative strategies.

To quote one of the speakers,

“Data is a core problem with ongoing and massive potential for impact.”