Coca-Cola overcomes challenges to seize BI opportunities

Imagine integrating data from 74 unique databases, all using distinct software to store and analyze data. Imagine trying to stitch together this patchwork of homegrown and small-scale analytics systems into a unified platform that could enable a large national manufacturer to quickly generate sales reports, manage production output and identify when retailers will need to be resupplied.

That was the problem soft drink giant The Coca-Cola Co. faced in 2010. The beverage products that end up on store shelves across the country are manufactured by a nationwide network of 74 bottlers, each of which has exclusive distribution rights within certain territories. Some of the bottlers are much smaller than others; they often were using very simple systems to analyze their business operations, in some cases just Excel spreadsheets. At the same time, other plants had developed homegrown analytics systems that didn't uniformly generate reports in ways that could be understood by Coca-Cola's corporate systems.

Speaking at the 2013 TDWI BI Executive Summit in San Diego, Justin Honaman, a former vice president and director of customer intelligence at bottling operation Coca-Cola Refreshments, said dealing with the difficulties associated with disparate databases and software systems was the chief problem during his time at the company, which lasted from 2003 until spring 2013. Internal analysts at Coca-Cola knew there were BI opportunities in the mountains of data the bottlers were collectively storing, but finding and accessing all of that information for analytics uses proved nearly impossible.

"To get to an answer, you have to have all the data together," said Honaman, who now heads the U.S. consumer goods consulting practice at data warehouse software vendor Teradata Inc.

The disparate sources of data resulted in long lead times in getting analytics reports from IT to Coca-Cola's sales teams. They also saddled the company with an inability to analyze large volumes of data and disconnected IT systems across the organization, Honaman said. In order to solve those problems, the company decided to rip out the legacy software at every bottling facility and standardize each one on a new business intelligence system, a combination of MicroStrategy and Microsoft BI products.

That may sound like a major technical undertaking, but Honaman said the challenges were more organizational than technical. Initially, business users had a hard time adjusting to the new tools and weren't using them effectively.

"We thought we were giving them the shiny, top-of-the-Gartner-BI-Magic-Quadrant systems, and what it turned out was people missed the old system," he said. In particular, users complained that the MicroStrategy software had too many options, and they didn't know what to do with it. 

Many IT departments and BI teams are familiar with users resisting changes to enterprise systems that have been in place for years. Once people become familiar with applications, it's not surprising that they wouldn't want to see them changed. But Honaman said there are ways to make the rollout of a new BI and analytics system smoother.

For example, Honaman said he should have asked representatives from the operations, marketing and sales departments, who would be the primary users of the new BI system, to be more involved from the beginning. The users mainly objected to the way reports looked and how they had to interact with the system. But those are things that could have been changed at an early stage. The IT department simply didn't know what the users on the business side wanted.

Additionally, IT and BI managers should be prepared to talk about the benefits of an analytics system in everyday business language. As soon as Honaman started talking to Coca-Cola's chief marketing officer and other members of the marketing team about databases and file systems, they would stop paying attention. But whenever he talked in concrete terms about how a unified system would help the company track customer sentiment, clickstream data and social media brand engagement, the marketing team would become more engaged.

"I think this is an opportunity for all of us. Drop the technical verbiage," Honaman said.

Ed Burns is site editor of SearchBusinessAnalytics. Email him at [email protected] and follow him on Twitter: @EdBurnsTT.