6 big data dos and don’ts

Research reveals best and worst practices for analyzing big data

Executives are changing their minds about big data. Increasingly, more organizational leaders are recognizing the importance of strategically capturing and analyzing data for a wide array of reasons.

For many organizations, this process quickly turns into an overwhelming exercise. With access to newer data sources like streaming device data, unstructured social media data and more levels of online transactional data, many organizations don’t know where to start looking for answers, let alone how to ask the right questions.  

According to an IDG Research Services study, only 26 percent of respondents say their organizations are capable of knowing which questions to ask. But maybe that’s OK. If you let the data guide you, the questions can become obvious.

“Those leading the space are starting with the data and letting the data help drive the organization to the question,” says Scott Chastain, Senior Systems Engineer for SAS. “It’s an old-school approach to start with a question and then go find the data to answer it. That’s a significant flip.”

What other advice can we gather from SAS experts and recent IDG research? Read on to learn six dos and don’ts for approaching analytics projects with big data and unstructured data sources.

1. Don’t assume the most ambitious approach will garner the best return. While a big payoff is tempting, it’s dangerous when businesses embrace unstructured data by trying to do everything at once. Instead, organizations should look for the smaller, easier-to measure applications as pilot opportunities, and then build momentum off initial successes. Picking the right projects to test capabilities is crucial.
2. Don’t focus efforts exclusively on business unit needs. Success is most achievable when organizations think globally, yet act locally. “Decreasing measurable risk is often more effective as an initial project than focusing on better understanding competitors or creating new opportunities,” says Fiona McNeill, Global Product Marketing Manager for SAS. “The latter are often significantly harder tasks, and they take longer to measure. The goal is to find projects with immediately measurable ROI.”
3. Don’t expect technology alone to guarantee desired results. Although the survey shows the lack of technology being a top obstacle – especially among small and midsize businesses – the technology is available. And, in some instances, organizations can leverage open source solutions or proven solutions for free to test their capabilities on a trial basis. There is no silver bullet when it comes to data success. Unless it’s an analytics-driven company, the capability and focus become less important, and achieving ROI is the goal.
4. Do build collaborative capability. Using toolsets and processes that make big data approachable often proves instrumental in solving some of the major challenges outlined in the study, explains Chastain. “When looking at cost and skill set, big data hype is growing faster than the capabilities. You need to create an environment that facilitates ease of use,” he says. “The ability to harness the power of big data requires more than technology. It requires business and IT collaboration. Organizations that foster the most collaboration will be the ones that benefit the most – and the quickest – from a Hadoop platform.”
5. Do take a stepwise approach. Rather than making the common mistake of trying to tackle unknown problems with unknown data, the most successful organizations start by solving a known problem in a new way. The next step is to then solve the same problem with new data, and then move on to solving new problems with new data. “The organizations taking a stepwise approach have the highest likelihood to achieve success,” says Chastain. “For instance, if a mobile telco has an established means of measuring churn, big data makes it feasible to see if social media yields improvements because it is a matter of addressing an existing problem with new data.”
6. Do think strategically, yet act tactically. As organizations embrace big data, they are often focused on building a platform to solve a specific business problem. As a result, the program is typically seen as an experiment, which makes it difficult to evolve and integrate as a business or enterprise asset. However, with strategic goals in place, it’s easier to see the second, third and fourth applications. This is important because it’s often the continued applications for which businesses realize big data’s true value and potential.

There is no silver bullet when it comes to big data analytics, but success starts with a solid strategy. We hope you can use these tips to glean valuable insights ranging from process optimization to customer-facing improvements.