This article in Internet Week points out that data quality is a serious problem in companies seeking benefits through e-procurement. Data inaccuracy not only hinders realization of benefits, but it hinders the amount of spending analysis that can be done to estimate benefits in the first place.
I have observed for years that the quality of data in many purchasing systems is poor, with duplicate vendor numbers, obsolete items, incorrect category codes, and so forth. The problem is often worse on indirect items, where many companies begin their e-procurement efforts. As long as informal, paper-based systems are the norm, buyers know how to work around data inaccuracies. But once the system goes electronic, all the data issues become serious impediments. This article recommends that companies apply serious data quality efforts to purchasing data that are similar to those used in data warehousing, such as data extraction, cleansing, and normalization. Although I don't think most companies will need to go this far, it might be warranted in companies doing many acquisitions, where the acquired companies need to merge existing data into a central purchasing system.
