Big data is a “top of minds” topic nowadays. CEOs across the world read magazines on planes and ask their IT leaders whether they should invest in big data. Should we adopt Hadoop? Apache Spark? Data Torrent? Any CIO grappling with these questions should first go back to the core question: What business outcome are we trying to achieve?
I remember coming into my role as CIO for FrieslandCampina Asia about four years ago. Management meetings were pretty much “paper based” and lots of time was spent arguing the data. So we set out on a business intelligence journey, automating data collection and presenting it in highly visual dashboards. Instigating processes where data was published a few days before “The Meeting” so managers could argue the data upfront, but not in the meeting itself. We rolled out sales performance dashboards, financial performance dashboards, and a few more.
Until one day, our Asia CEO told me the dashboards were all good and fine, but he did not feel we were moving the needle in terms of top and bottom line business results. I remember him saying “I don’t want just dashboards; I want your people to walk into the Sales Director’s office on Monday morning 9 am and tell them what to do!”
As of that moment, our data science journey was born. We hired our first Data Scientist. I asked him to validate our Go-To-Market strategy for the General Trade. About 60 percent of our revenue in Asia comes from High Frequency Stores (sometimes called “mom & pop” stores), which are visited by our own sales reps who carry handheld devices to record sales. We record daily sales for each product in each individual store. A wealth of data we never “mined”.
"Putting in place digitization programs to automate commercial processes and collect consumer and shopper data is fundamental for continued success"
Quickly it became clear we had opportunities to improve: which stores can we visit profitably, what assortment is best for each store, what consumer promotions work or don’t work etcetera. Initially the methods and algorithms were simple: K-means clustering, pare to analysis, statistical correlations. Simple methods, but the business impact was very significant.
We decided to expand the data scientist team and started looking for other areas to drive business value— pricing and promotion optimization, distribution across the region, cross border sales analytics for specific countries, and online consumer behavior.
At this point we decided to invest in a big data platform: some analytics get more efficient with a big data platform, some analytics cannot be done with traditional data bases. Especially when you want to analyze consumer behavior, big data is quickly required. Another angle to this is to collect more relevant data. Digitizing processes, especially in the consumer space will provide more data to mine. Putting in place digitization programs to automate commercial processes and collect consumer and shopper data is fundamental for continued success.
I realize nothing of this is “new”; I merely wanted to emphasize the journey. Big data is a means to an end, and the end needs to be clear before investing. Our initial “small data” analytics led to a region wide program to revisit our Go-To-Market strategy. Plus it led to a belief in data science, allowing us to invest in both people and big data technology.
Big data is undoubtedly the future and we will continue our journey to provide data driven business insights to our business counterparts. My plea is to make this a business, not technology driven initiative: focus on the desired outcomes, start small and once proven, big data will follow.