Start Making Molehills from a Mountain of Big Data

September 10, 2014

molehill_or_mountain.png

Original photo by Paxson Woelber, modified for this post.

This post is a part of a series from the partners at Datascope, titled
Business Strategy in Focus: Bring Me My Monocle

Have you ever known someone who outfits themselves for a camping weekend as though it’s a month-long safari when all they really need is a small pack with the bare essentials? It’s tempting to take on a big challenge by preparing for every possible contingency, but this isn’t the only way to address major challenges. Rather than making a mountain out of a molehill, we advocate a different approach.

Problems at large corporations—especially as they involve emerging fields like “big data”—take on a larger-than-life importance that seemingly require equally big and complicated solutions. But adding complexity for its own sake rarely pays off. It tends to require more resources than necessary and take longer to achieve results without a guarantee of addressing any real issues. From our perspective, this approach to solving problems (“we need a data warehouse before we can do anything!”) is not unlike our friend with a 100 pound pack for a weekend camping jaunt.

Break Down “Big Data” into Simple and Manageable Data

How do you climb a mountain? One step at a time. Every large challenge can be broken down into smaller, more manageable issues that make giant tasks easier to tackle. Rather than trying to solve the looming question “What is our data strategy?” with a complicated and expensive answer, break the question down into an assortment of smaller and more concrete questions. This way, the questions are easier to answer, with solutions that are more tractable to implement.

The key benefit of this strategy is that it presents a visible path to ultimately overcome the original, larger challenge. It accomplishes this by asserting a different set of questions that leads to identifying a better goal, and subsequently by uncovering a smarter approach to solve the original problem.

Data Scientists Ask the Questions You Can’t

In examining any business issue, a good data scientist will start without any assumptions, and will ask some elementary but critical questions - not like a naive toddler, but like an empathetic and inquisitive professional:

·      Why do you have this problem?

·      Why do you need to fix it?

·      Why did you identify this problem?

Framing the issue through curiosity is the key to making sure you’re working toward solving the right problem. It is a simple but powerful technique that data scientists use to question assumptions and reveal new perspectives.

An Example: Using Customer Feedback Data to Improve a Manufacturing Process

Motorola Mobility wanted to improve the management of their supply chain and manufacturing processes by better incorporating customer feedback. The challenge they initially identified was that by the time they received customer feedback from third-party retailers that sell their devices, it was often too late to investigate and fix upstream quality issues, as they were already manufacturing a new model.

Rather than trying to change their manufacturing and supply practices or pressure their retail partners into being more prompt with handing over feedback, we suggested a different approach and asked a new question: “What other sources can provide more timely feedback?” The answer to this question opened doors to alternative solutions that were less complex, less costly and yielded faster results.

Asking Questions Improves Perspective on the Hill to Climb

In 2006, Netflix offered a $1 million award to anyone who could figure out a way to improve their viewing recommendations by 10%. It took competing teams nearly three years to reach this goal. In the end, Netflix gathered some highly sophisticated algorithms, but despite huge potential benefits, the winning solutions were too complex to actually put into production.

If Netflix had framed the problem differently, i.e., “we want to improve recommendations” rather than “we want to improve our recommendation algorithm,” the answer might have been easier to come by. As it turned out, Netflix ultimately found a much easier answer: splitting out accounts to accommodate the preferences of multiple family members and to provide more personal recommendations for each user. We think they were probably just a good question or two away from starting on this journey directly, and avoiding the more circuitous route.

The Best Data Solutions Might Not Be Big Solutions

Big companies tend to gravitate to big solutions, especially when it comes to data. Yet there is greater value exploring the new opportunities afforded by data in the form of small challenges. Try breaking down a problem to its fundamentals. We find that this process can reveal root causes that offer a new point of view on the scope or imperative of the ‘big’ problem. Not every problem is as big as it seems, nor is solving every big problem the answer to yielding an equally big improvement or result. Unpack the baggage that covers every possibility, and prepare to overcome the molehills instead.

contributors to this post

headshot of Datascope Partners
Datascope
headshot of Bo Peng
Bo
headshot of Brian Lange
Brian