Last Saturday, the Interanians crew headed down to LA for Big Data Day LA. Almost 1500 people showed up on sweltering Saturday afternoon. What I noticed right away is how much the tech scene has grown in the City of Angels. The top 100 tech companies employ over 20K technology professionals.
The underlying sentiment was that Excel and SQL are no longer sufficient to even handle the scale of today’s average business. Many companies accrue technical debt at a high rate, especially when building outside of their core area. As a result, big data projects can end up with more questions than answers. Engineers tend to be curious about the technology behind modern storage and analytics, but decisions makers want fast, accurate results rather than a science experiment. Corporate culture is also closing the loop between analytics and product development and requires faster answers to support continuous integration.
I presented in the Big Data Track. My talk was entitled “Purée through trillions of clicks in seconds”. I picked that title to roll the intersection of data and rapid iteration into a single source of truth. Analytics are more like purée: you have to take tons of disparate data source together and make them easy to slurp quickly.
Difference Between Analytics and BI
One common misconception is that Business Intelligence (BI) and Analytics tools are interchangeable. They definitely have overlap but Analytics requires the ability to focus on true behavior. For example, consider tracking Daily Active Users (DAU), Monthly Active Users (MAU), and Conversion rates. These are top line numbers that aren’t immediately actionable. Analytics instead would ask “What’s the dropout rate when errors occur within the first 10 minutes of a session?”. This looks at true behavior and is actionable from a product point of view: we can root cause the errors and reduce dropouts. BI and Analytics tools require different architectures due to the depth of questions. BI systems rely on aggregation and heavy caching along the temporal axis. Analytics requires fast scanning and organization to match the first-order actors in the system.
Advocacy and education helps the community at large. This can start from grass roots with Meetups and even institutional knowledge. Analytics 401 maybe?
Thanks to Kim, Matt, Hector for warming me up and help setting up the event. Special thanks to Miroslav for providing me content for the presentation and a live demo site!
If you’re interested in learning more about behavioral analytics, you can visit our resource page for details on e-books, use cases, and videos.