Systems integration is the problem nobody wants and everybody has. It's about getting all of your computer programs to speak a common language so they can relay information back and forth. Consider how much more efficient your supply chain would be if your order management system could get a look inside your warehousing system, and vice versa. But it's not just about internal integration. Often you need your computers to talk to other computers that reside with a supplier or customer.
Many agree that the problems with systems integration are more about budgets, manpower and corporate culture than code. "Integration is the Number One enemy of projects—the main reason for project cost overruns and going over schedule," laments John Fontanella, who is research director at AMR Research Inc. in Boston and a specialist in supply chain issues. "Application vendors rightly say they can implement their application in three months, but the typical time is six to 12 months because the client company can't free up the IT resources necessary to weave it into the overall ongoing business."
Plus, integration often costs far more than the software itself. "It's a wild card. With the average ERP (enterprise resource planning) project, the license fees pale in comparison with integration and implementation costs, which usually [run] five times more," Fontanella warns.
The irony is clear, says Fontanella. Though costly, "integration is the key to unlocking the value of all these programs." i2's supply chain management software, for example, is designed to suck every bit of information from a company's operations systems in order to figure out the most efficient way to run those operations in concert with one another. "Unfortunately that never happens, so a lot of the expected benefits are never realized owing to lack of integration," he says. "And companies become tired of all this stuff. They get fed up."
No more tangles
To side step those integration hassles, some companies simply choose to buy all their supply chain operations software from a single vendor. That way, they're guaranteed (in theory ) to have software programs that swap data smoothly. This was the approach taken by Fingerhut Direct Marketing Inc., based in Minnetonka, Minn., which sells hundreds of millions of dollars worth of goods ranging from lawn ornaments to leather jackets via catalogs and the Internet. Fingerhut got a new lease on life when it was bought by two entrepreneurs last July. It also got a fresh perspective on the tangle of legacy systems that was controlling order management, merchandising and warehousing.
That perspective prompted Loren Eggert, vice president of operations at Fingerhut Direct, to ditch the whole lot and start afresh. His team picked HighJump Software of Eden Prairie, Minn., to install its warehouse management system (WMS) and supply chain visibility systems, linked to a Microsoft Great Plains host system.
The work had to be done fast—in time for Fingerhut's Christmas season, which accounts for over half of its annual sales—so there was a nail-biting period when Eggert worried whet her High Jump could live up to its promises. But, some teething problems aside, it all worked out. "The biggest thing I've learned is to have open, face-to-face communication,"says Eggert. "And walk before you run—a phased-in approach is a sound approach. We didn't put in the Cadillac version right away," he says. "We put in the Volkswagen version initially, to get speed-to-operation for the fall season. I' ll have my Cadillac version by mid-summer."
Of course, there were sacrifices to be made. Fingerhut opted for a phased-in approach, installing only the core competencies of the warehouse management and supply chain visibility systems at first. That meant there was a two month period after the legacy systems had been shut down when there was only limited capability in the new system. Eggert reckons the company's operations experienced about a 30-percent drop in productivity for those two months. But, he says, the pain was worth it.
Apart from improved internal integration, Fingerhut has experienced another benefit from running software packages that use a common language (SQL).The company sells some third-party fulfillment services, which involves running interfaces between Fingerhut's order fulfillment software and a client's system. Before the changeover to integrated operations, it could take up to six months to successfully hook up the two systems and get the right data flowing. Now, Eggert says, it takes 30 to 60 days. Another notable change is that corporate Fingerhut has reduced its IT staff from around 350 to fewer than 50.
Why don't more companies take this route? Despite the obvious advantages of choosing a common platform, says Fontanella, "not many senior managers have a true appreciation of the benefit of electronically connecting all the parts of your business, so it doesn't happen very often."
There are other obstacles, too. "For a company of any size to standardize on a single system takes incredible patience and an extraordinarily strong vision. This can get very political inside an organization," Fontanella adds. Another drawback to using only one application vendor's multiple modules (warehousing, order management, transportation management, etc.) is that you can't guarantee that you're getting the strongest product in each individual area. "You have to be consciously dedicated to the one system because you need an integrated enterprise, and you have to be prepared to forego superior functionality for that," Fontanella says .
However, he points to various large companies, including Colgate— which runs 85 percent of its business applications using products from German software giant SAP—that are really making the single-vendor strategy work. "They can do analytics and absorb information from all over the enterprise," Fontanella says.
Compromising positions
Some, like Colgate, rely almost totally on a single vendor. At the other extreme are the highly decentralized corporations that act almost like a holding company, consisting of autonomous units that report centrally. Those tend to basically have a data warehouse with no systems integration.
Then there's the vast middle ground, where the majority of companies fall. Most companies tend to compromise, buying specialized logistics management products from a small vendor and more general supply chain software from bigger companies like Manugistics or SAP, t hen trying to weave them together, explains Kimberley Knickle, also a research director at AMR Research and a specialist in general computer integration issues.
"When we talk about integration in supply chain software, you have to step back and ask yourself what you're trying to do," says Knickle. A company might be trying to achieve visibility, for example, or it might be unveiling a system that lets customers serve themselves at a Web site. Or it might be aiming for collaboration with a client or supplier company. "You have to figure out the motivation and, after that, start thinking about specific tools. Is it about people? Is it real time or do you just want an update once a day or once a week?"
Knickle says the answers to these questions could determine whether you end up with an Internet-based pOréal for gathering and exchanging information; or a connective web of middleware that allows deep integration with a customer's or client's back-end system; or a simpler EDI (electronic data interchange) connection through a VAN (value added network).
Whatever the choice, a combination of solutions tends to work best, says Dave Adams, vice president of global operations at GT Nexus in Alameda, Calif., who has spent the last two years helping the Web pOréal company make direct back-end to back-end computer connections between shipping companies representing 40 percent of the world's shipping container capacity and their customers.
In an ideal world, he says,all companies would be integrated back-end to back-end, but that's not always practical. Often, the gaps can be sensibly filled by using Web-page information entry. That means building a Web site, usually one that's password protected and easily accessible through the Internet, where people can go and enter information. The manual entry increases the risk of error and takes up more time, but it's often the best solution when hooking up computers directly would be too difficult or expensive. "A lot of people tend to oversimplify integration. They say: 'Let's connect the systems and let them talk and we'll all be happy ! 'That's great, until you start counting the number of systems on the other side of the fence. You'll say to a carrier: 'I want to integrate with your booking system,'and it says: 'OK, we have 23, which would you like to integrate with?'"
Adams says the trick is to be prepared to compromise on a solution that includes some fully integrated operations but also some patches with Web pages. You also need to accept that some customers or clients will be able to hook into your beautiful new system and some will want to rely on Web entries for a while longer. Or flat files. Or even faxes and phone calls. Almost inevitably, Adams warns, "[y]ou end up with a hybrid integration model."
Get real
Adams, who has done more integrations than most of us have had hot dinners, has some other guidelines for shippers looking to get all of their supply chain software onto one system (even if it includes software from different vendors). He advises any company to make sure the internal IT staffers actually working on the implementation are given realistic workloads. "There are usually only a handful of people who are doing this, maybe six to 20 people, and there's always pressure on their time," he says. Adams points out that although a CIO may be eager to hook up to an ocean pOréal such as GT Nexus and get his IT team working on that,a huge customer could come along at any time and demand that the shipper deal with it on its own chosen computer platform. "If Wal-Mart or Nike wants to hook up, the IT team has a tendency to drop everything to make that happen," Adams warns. In other words, make sure you talk regularly to your tech people about how they are prioritizing integration work.
Adams agrees with Fingerhut's Eggert on the importance of regular communication among everyone involved. "Eighty percent of the battle is expectation-setting up front," Adams says. "We rarely find a situation where the person on the partner's side is incompetent and doesn't know how to set up a protocol or whatever. If things go south, it's because I didn't understand this was a priority or 'I'm in Korea and you're in NewYork and we spent five days playing phone tag.'"
Finally, it's a good idea to clean house before you start messing with your computer operations. That means getting databases corrected for errors, as well as standardizing data entry. Adams says that simply compiling a common customer database is a challenge for many international companies, which might have six different ways of entering "IBM" depending on geographical region or department. "Technically, integration is not difficult," Adams says. "But a lot of customers need to do a big internal data cleanup before they can do this well."
