What you need to make Big Data work: The pencil
Building out Big Data capabilities too often becomes the end goal itself.
- October 2012
- Matt Ariker
The secret to getting the most from Big Data isn’t found in huge server farms or massive parallel computing or in-memory algorithms. Instead, it’s in the almighty pencil.
That’s the advice I give people these days. The reason is that the promise of Big Data is so seductive that it often sends people scrambling after ever more petabytes or exabytes in the hopes of unearthing that golden insight that will allow them to grow and beat competition. This process untethers many a marketer as one question leads to another and to another. Many retail and telecommunication companies I’ve worked with, for example, have sought to create a 360 view of the customer by spending years building out an enterprise data warehouse to support it before generating new revenue streams. The build-out itself becomes the end product, not the analytics and revenue impact.
Don’t blame the data. Tremendous insights do exist in Big Data. Companies that use it well are leaping ahead of their competitors. One of the big reasons for that, however, is that they have a very clear sense of what they want to do with all that data before they start.
Which brings me back to that pencil. It’s a simple but powerful tool to evade the Big Data trap of analysis paralysis. Here’s how:
1. Start with “Destination thinking”
Write down in short, clear sentences exactly what business impact you want to achieve with your new Big Data analytics. This “Destination thinking” is a simple but often overlooked process that goes beyond expressing broad goals such as “increase wallet share.” You want to lay out what business questions or problems you expect to be able to solve when you have finished the analysis.
My colleagues highlighted a good question in their recent piece in HBR (“Making advanced analytics work for you”: What decisions could we make if we had all the information we need? Using that logic, one shipping company improved the on-time performance of its fleet by tapping specialized weather forecast data and live information about port availability that it hadn’t realized were available. The very act of writing at this level of specificity will help you clarify what you’re looking to do, and how you will define and declare success before you start. I can’t tell you how often people have done this and told me how it’s been instrumental in helping them come to a much better understanding of what they needed to do. It will also help to drive alignment with your team and bosses, a great way to defang ambiguity, the serial killer of Big Data initiatives.
2. Define what success looks like by digging into the nitty gritty
Set hard and measurable goals. Lots of folks talk about improving customer experience, a worthy goal. But how do you measure it? Goals like revenue growth, increased profitability, or increased customer use are good because they’re measurable. But don’t stop there. Write out how the improvement you’re shooting for will impact the P&L. For example, if you want customers to stay longer, are you expecting them to increase their product usage too? If you’re looking to reduce customer churn, how much of a reduction, for how long, and how much stronger will profit be because of it? Test the metrics you’ve laid out: for these measurements, how will you ascribe causality and measure it?
3. Define milestones of success and early opportunities to generate business returns
Big Data analytics initiatives tend to have long gestation periods so it’s important to identify milestones that need to be completed. Focus on what I call “Insight Delivery Deadlines.” Installing necessary hardware and software infrastructure, for example, is important but in and of itself doesn’t deliver value. A better Insight milestone is “convert first high-end target into customer” or “deliver insights report to CEO.” Just as importantly, take that pencil and put people’s names against the various milestones so that everyone is clear who needs to deliver what and when throughout the entire Big Data gestation lifecycle. Make sure your team is clear and accountable about these milestones (again, kill that ambiguity).
In completing these steps, you’ll create the requirements for the insights from Big Data analytics you actually need versus what you think you want. Of course there’s a lot more to do — for example, you’ll need to establish a baseline so you can measure improvement. But putting this kind of time into your Big Data project upfront will help not only save lots of time and money during implementation, but you’re also much more likely to get the value you’re looking for.
So, step away from the Big Data hype for a moment and grab yourself a trusty pencil.
Submit a comment
Comments chosen to be published may be edited for length and clarity and will appear along with your name and details, but not your e-mail address. We will use your e-mail address only to send you a confirmation copy of your comments and to notify you if we publish them online. We value your feedback and will consider it carefully. Nonetheless, we receive so many comments that we cannot acknowledge all of them.