How to achieve scale in analytics
Blog Posts:Mu Sigma
Published On: 29 January 2015
In previous blogs, we’ve discussed how to successfully blend the art and science of big data analytics. Today, we tackle how to achieve scale.
The pace of innovation in the world is accelerating rapidly. The number of major breakthroughs in the past 100 years may equal or surpass the number made in the previous 1000 years. Scale has a lot to do with this.
Scale is difficult though when it comes to problem solving and analytics. You can’t scale analytics with assembly line improvements or quantity discounts. For analytics to truly scale up, there needs to be a focused attempt to separate out the obvious from the exceptional, and a sustainable, structured process in place to help identify problem components and patterns.
Consider an ice cream sundae. The ice cream itself is mass manufactured and limited in variety – chocolate or vanilla, for instance. It forms the base of the dessert. The toppings can then be customized according to taste. In the end, you receive a concoction that’s a marvelous combination of standard and customized.
Likewise, while solving a problem one can scale the basic aspects of problem definition, solution framework, and the necessary analytical infrastructure. Overall, 80% of the analytics process can be performed in a structured, relatively simple and predictable way. The remaining top layer is all about the tailored add-ons one needs to arrive at the optimal solution.
There are standard approaches to untangling the knots and breaking down complex issues into smaller ones. Small answers can then be compared on a large scale. That will give you the patterns. From these patterns will emerge the solution to the problem.
Introducing scale into a data analytics platform sprouts two major sister branches. One branch will eventually reach out beyond the tree itself and explore new boundaries – fringe problems that might be loosely connected to the main problem. This will usher in newer opportunities, lead to expansion of the domain in which analytics might work, and eventually address the singular goal of growth. The other branch will end up touching the roots, strengthening the base of the tree – the ‘art of problem solving’ itself, allowing for further vertical growth or in other words, further scaling up.
It’s a continuous cycle of extensive, exploratory learning and feedback leading to improvement – the key to maintaining the art and science of problem solving in a state of continuum and it is this continuum that will keep the industry alive and thriving.
What’s worked in your organization in terms of scaling analytics efforts?