For commercial software vendors, 1997 will be a banner year. Five leading vendors-- Baan, J.D. Edwards, Oracle, PeopleSoft, and SAP--are expected to report combined applications- related revenues of more than $6.5 billion. Many smaller vendors are doing just fine, too. Year 2000 problems and the allure of Internet commerce are bringing in customers in droves.
Flushed with success, vendors are hyping their software's capabilities. In reality, however, packaged applications fall short in a number of respects.
Vertically inadequate. Many vendors promise "enterprisewide" support. But tell that to utility and telecommunications companies looking for new billing applications responsive to their changing markets. Most are obliged to turn to customized (and expensive) solutions from systems integrators.
Vendors claim to support financial services markets, ignoring the fact that there are eight SIC codes under the insurance segment alone--each with unique needs. Even in manufacturing markets, which have historically benefited from broader coverage, no single vendor can deliver more than 50 percent of the necessary applications portfolio. Product design, logistics management, shop-floor control--most processes outside headquarters are still serviced by specialist vendors.
"Good enough"--not best--practices. The vendors trumpet, "Reengineering in a box." But ask them what the average cost of invoice processing is at their user sites, compared with world-class benchmarks. Few know--or care.
Many vendors sell their Web-enabled "front end" as proof of their support for electronic commerce. But good luck finding many packages that support collaborative forecasting or schedule sharing with trading partners. Other unsupported business practices include balanced-scorecard reporting, cross-functional business forecasting, yield management, and management of contingent labor.
Regional, yes--global, no. In a recent Price Waterhouse survey of U.S. software companies, only 12 percent offered Japanese translations, and just 6 percent offered Italian translations. European vendors (and U.S. exceptions such as J.D. Edwards) do somewhat better on geographic coverage. Indeed, there are a number of SAP-based, pan-European shared services projects. But many primarily focus on processing payables. Common customer-order or payroll processing is still a rarity.
Meanwhile, the Asian corridor, with its double- byte requirements and lack of a common market, challenges vendors. Latin America and Africa present other issues. The bottom line: regional software projects are tough enough; global homogenization attempts are still too ambitious.
Rapid implementation? Not so fast! Vendors claim that an "average" implementation will cost no more than twice what users pay for the software. But such accounting is inconsistent. Many do not impute the cost of user staff in such estimates. Others focus primarily on the cost of installing and configuring the software. Most applications vendors still do not offer tools that make integration with legacy systems, reconciliation of converted data, stress testing, or end-user training significantly less labor intensive.
Future shocks. Change is constant in the client/ server world. Two years ago, Oracle delivered a fat "smart client"; now, it's going back to a thin client. Users have little choice but to follow vendors. PeopleSoft, for example, will not support older releases past an 18-month window.
An upgrade is like refueling in midair: you cannot shut down a production system for longer than a weekend. Companies must rehearse the "cutover" several times and test customizations thoroughly. One multinational recently reported it spent $5 million to upgrade from SAP R/3 Release 2.2 to Release 3.0--with no new functionality implemented.
My recommendation? Plan for a major upgrade every two years, and budget 10 percent of the original implementation cost, each time.