If you weren’t able to join us at Strata+Hadoop in New York City last month to hear how P&G uses on-cluster BI to deliver visual insights to hundreds of business users for everyday use, you’re in luck. Doug Black, Managing Editor at Enterprise Tech, was there and recently published an indepth article based on the session.
It’s not everyday that you get an inside look at the workings of a $76 billion dollar CPG company. But every day is something Procter & Gamble knows well. Terry McFadden, principal enterprise information architect at P&G, shared the ins and outs of his multi-year big data POC journey to bring together dozens of datasets to iterate actionable insights. Legacy tools just weren’t going to solve their needs, as quoted in the article:
“Traditional approaches for gaining insight from data just weren’t adequate,” McFadden said. “The volumes were growing, the costs were growing, and the time-to-insight required in terms of the business event cycle was speeding up, and so this is a real challenge.”
When thinking about what they were looking for with regard to building their architecture, McFadden knew early on they wanted an analytics ecosystem that didn’t take their data off-platform. P&G chose the Cloudera Hadoop distribution which brought to their ecosystem the performance and power of Cloudera’s Analytic Database. After this decision, they started to explore BI and visualization options for best suited to the architecture.
“We believe fundamentally in the model of moving the work to the data, and not moving the data off platform to the work,” McFadden shared. “Those that move the work to the data continues to show an ROI that allows you to build out more and more.”
We couldn’t agree more. In fact so much so that we architected Arcadia’s Native Hadoop BI platform to bring the BI processing to the data in Hadoop. McFadden calls this “on-cluster BI” in his talk and it is the only way that makes sense with growing data volumes and user demand.
Another key takeaway from the talk centers around gathering use cases from business users. It’s one thing to have access to the data, it’s another to know what questions to ask of it. The article relays how McFadden’s team sent embedded analysts a survey asking them to: “Imagine you had a magic 8-ball and you could ask it anything, what are the wicked problems you wish it would answer.” The replies helped provide use cases to guide the initial proof points for McFadden’s team.
There were lots of specific questions from the audience after the presentation to dig into the details of Procter & Gamble’s implementation. Even after the session was over, McFadden found himself engulfed in a crowd eager to learn from his experience. Just like with big data, it’s about what questions can be answered and what answers should be questioned. This week, we all know the answer is: Go, Cubbies!