Getting the business foundations right first before diving into Big Data.
According to Gartner, one of the world’s leading information technology research and advisory companies;
“Big data can help address a wide range of business problems across many industries and for the third year in our study, both enhancing the customer experience and improving process efficiency are the top areas to address,". (http://www.gartner.com/newsroom/id/2848718).
Generally, customer experience is a huge part of what organizations are most keen to obtain greater insight into, particularly in the business services area. Manufacturing can certainly gain from greater efficiencies which big data can also provide. However according to Dexter E. Wood Jr., Senior. Vice President and Global Head Business and Investment Analytics, Hilton Worldwide, it is important to do the business side of things first and deal with the data after, he says, “What are all the metric definitions going to be, how are we going to use the data?...define your business rules, define the metrics and then go find the data to support those” Dexter adds, (YouTube, 2014).
According to Dave Clarke, A Data Scientist at Asystec Data Management Solutions, as reported by Ian Campbell in The Sunday Business Post, (November 2014), they (Asystec) start their customer relationship process with a two day workshop where businesses ask themselves the questions they want answered, Asystec then take the results into their data science labs for between six and twelve weeks.
“The objective is to identify key performance indicators that help answer the questions. Then Asystec provides a cloud-based analytics solution that can begin to query the data. It uses the EMC Greenplum Platform with ‘R’ the open source programming language that is increasingly used to analyze big data and surface answers to the specific questions that organizations want answered”.
It’s often, as Asystec suggest, best to start with the basics around the structured data that the company has and only when your model is working should you look at bringing in unstructured data like social media.
Wikibon Image shows the importance of model iteration.
Intercepting Twitter feeds for company or brand feedback is something many companies are experimenting with according to Ian Kennedy Compston, pivotal solutions specialist with Triangle. According to Niamh Townsend, enterprise solutions director at Dell, she advocates that an organization starts with one departments problem or question on internal data and build on from there to other departments and finally bringing in none structured data from outside, (Sunday Business Post, November 2014). All these writers suggest incremental stages of data management and a gradual approach to incorporating Big Data into an organizations data retrieval and insights strategy.
Finally, apart from clearly defining your research questions you must also ensure that the right people are given access to the right data. Jason Burns, software client architect at IBM says that the whole enterprise must sign up to data management and strong governance level management of data retrieval is necessary before proceeding (The Sunday Business Post, November 2014).