7 Dimensions of Big Data Agile Analytics

Updated from 2 April 2014

Agile Analytics is a blend of sophisticated analytics techniques, agile delivery methods and lean learning principles, but doing it well involves a bit more than that. I recently gave a talk in Singapore where I outlined the essential elements of effective agile analytics. I summarize these seven dimensions here, which accompanies this slide deck from my original talk.

Big Data. While I've grown cautious of this buzzword, it is important to acknowledge that today’s data is more diverse and messy than ever before. Agile analytics involves the use of both SQL (relational) and NoSQL technologies to create a polyglot persistence infrastructure for data management. The ability to align your data storage technology to the nature of your data is key to flexibility and speed.

Advanced Analytics. Creating true value from data requires going well beyond conventional BI reporting, which is important but rearward facing. Competitive disruption comes from predictive and prescriptive analytics (Data Science) and visual storytelling.

Agile Discovery. Agile analytics starts with a high value business goal and then chunks up the data science into tiny incremental goals that can be presented to stakeholders every few days. Each tiny analytical discovery can be reviewed and used by business decision makers and is a step toward achieving the bigger original goal.

Lean Learning. Each tiny business action or decision based on analysis is rigorously measured so that we can learn whether the analytical model has the intended consequence. This build-measure-learn cycle is key to ensuring that we are converging on the right outcome even if it diverges from the original goal.

Impact. Agile analytics is continuously focused on actionable, insightful, and disruptive knowledge discovery. It is very easy to find yourself doing interesting data analysis that is not necessarily impactful analysis. Delivering results every few days enables us to continuously evaluate the usefulness and impactfulness of the results.

Solutions Thinking. The best analytics are not simple answers to singular questions. Rather they are the complex heart of powerful solutions to challenging business problems. Examples like using machine learning on traffic flow data in real-time to reduce gridlock and traffic jams or using dynamic, real-time bioinformatics embedded in a sports bra to detect breast cancer in women are at the pinnacle of such solutions thinking.

Ethics. High ethical standards are needed to endure the inevitable scrutiny that analytics undergo. Standards include ensuring personal privacy and enabling individuals to control their own data. They also include radical transparency by the organization conducting the analysis. Democratic access to available data is key as is the use of open data standards among other considerations.

Consider your analytics programs and processes against these disciplines. How well are you creating frequent and continuous value from the data today? How excited are your business stakeholders about the disruptive things they can do because of analytics? How well are your analytical models and discoveries core to innovative solutions to complex business problems?

What can you do to get better?