What's Critical, What's Not

The flip pad on the easel in the corner of Berry's Haarlem office is down to its last sheet of paper. On that sheet is a diagram explaining how SOA works — red ink for the systems, green for the data flows, black for the names of the systems. And since explaining how SOA works and getting his staff to understand this new way of thinking (services, not applications) is one of Berry's hardest jobs, he hasn't let anyone touch the pad in more than a year.

Berry didn't realize what he was getting himself into until a two-day meeting that took place January 11-12, 2006, in a conference room at the Zoete Inval Hotel directly across the street from Coty's Haarlem office. Twenty-five key process owners, from functions including finance, customer service, distribution and IT, were going to create the integration plan. It was the first time that many of them had ever met.

"We are not a technology company," Berry told the audience. "We can't be writing code."

The point of an SOA, he continued, is to map the tasks that applications perform to the processes that a company follows in the course of doing business. At the technology level, these tasks — for instance, shipping an order — are translated from the proprietary language of the system that performs them into services written with XML-based standards. At the business process level, these services can be moved around and used over and over again to build new applications without ever having to write new code.

Berry told his audience that the purpose of the first day was to understand the processes that everyone used. Each of the representatives took turns talking about how they did their jobs and how they used their systems. What emerged was a picture of how the company worked.

The new company had five distribution centres. The two inherited from UCI — in Lille, France, and Mount Olive, New Jersey — ran on the JD Edwards platform. One Coty centre in Germany ran LM6, the other, Ratioplan. The North Carolina centre ran Oracle/COPs. They were all, says Berry, "old and tired". Furthermore, each country had unique processes (and in some cases, technologies) that had to be accounted for.

Italy, for instance, is already using the SAP system on which Coty will eventually standardize. The largest customers there usually place orders at the individual store level and expect products to be delivered directly to those stores. But England uses a legacy ERP system from JBA, and Coty's largest customer there, the Boots pharmacy chain, places orders by EDI and has them delivered to central warehouses.

There were smaller differences that the integration project would have to accommodate as well, such as label reformatting. The SOA project would have to identify all of these and turn them into universally readable services.

The day ended with a formal dinner and resumed on the morrow, focusing on how Coty was going to make all this happen. "We didn't have to build the Taj Mahal; we just needed a roof," says Gary Gallant, Coty's vice president of information management for the Americas, who came along in the UCI acquisition. "For example, we could have said anytime someone creates an SKU we need to replicate that in every system, but then we never would have made our go-live date." Instead, Coty limited its focus to anything that touched the customer: sending an order to the warehouse, ship notification, billing and so forth — what Gallant calls "the bare necessities of order management".

This focus on customer-facing processes came at the expense of internal ones. For instance, Coty's business leaders were used to receiving a daily report tracking sales and inventory levels. But Berry decided that integrating that report with the new UCI products would have to wait until after June 30. This was not a popular decision with the business leaders, but the daily reports, as nice as they were, did not directly affect customers.

In order to minimize the confusion, Berry decided to create small SOA project teams: one project manager, four IT directors and an iWay consultant in Europe, and an even smaller team in the United States. The countries that already had the SAP system in place would be integrated by Accenture, Coty's partner on the SAP project. Berry says that creating small teams was one of the best decisions he made.

"Too many brains can work against you," he says. "We didn't have the time to listen to every alternative." Smaller teams didn't require costly overhead like dedicated office space, and, most important, eliminated chains of command that might have prevented Berry from getting timely and candid feedback on whether "we were on the right or the wrong track".