ManageEngine’s Ram Vaidyanathan discusses the four key steps in executing an effective Big Data strategy.
In 2002, Billy Beane and Paul DePodesta famously used the principles of sabermetrics – in-depth analysis of in-game statistics – to assemble a competitive Oakland Athletics Major League Baseball team. They opted against signing big-name stars. Instead, they recruited cheaper, but statistically proven players. The A’s famously completed a 20-game winning streak, and finished first in the American League West.
If data analytics can have such a big impact in sport, it can undoubtedly do so in business where decisions are arguably less subjective. However, in today’s businesses, data is much more complex and vast than what DePodesta and Beane had to contend with.
The first step towards maximising business value with Big Data is obtaining it. For this, businesses need to extract both structured and unstructured data, and lots of it, from a range of sources. Sales reports, customer on-boarding dates, SLA breach rates, and web logs are examples of structured data. Social media posts, email messages, PDF files, and multimedia are examples of unstructured data. Unstructured and structured data each have their own strengths and weaknesses, so it is important to include both types in order to see the whole picture.
One example of pulling data from a range of sources is Duetto, a company that helps hotels personalise their prices by extracting and analysing historical data such as how much a guest typically spends at the bar or casino. Hotels can then incentivise guests with better room prices knowing that those guests will spend more on other services.
Once the data is extracted, it should be contextualised. Each facet of business should decide what metrics they will track and benchmark themselves on. For example, a sales leader will be most interested in tracking sales volume over time, sales volume by region, and purchase values of top customers. A marketing leader, on the other hand, may be most interested in metrics such as ROI, advertising reach, and profit margins over time.
After data has been extracted and contextualised, the analysis part comes in. First, analytics tools should alert users of any outliers and offer to exclude these outliers from the analysis. And, these tools should consider entire datasets, rather than sample subsets. Users should also be able to set up advanced techniques such as correlation analysis, regression analysis, conjoint analysis, and factor analysis, if required.
In the end, even contextually analysed data can’t do any good if it’s not being properly utilised. And when it comes to getting the most out of your data, it’s all about how you see it. Whatever big data analytics tool companies deploy should use data visualisation technology to help decision makers identify patterns.
While the challenge of monitoring so much data will remain, the new challenge will be to make Big Data analytics available to every employee. This will empower them to make smart decisions within their realm and create even more business value.