One of inevitabilities of the software world is the periodic rise and fall of expert systems. Whether it has been neural networks, learning systems, or inference-engines for classifier systems, these technologies have always held great promise--and usually great disappointment. In the procurement world, a vision of Web-crawling agents finding good deals and negotiating on buyers’ behalf has clearly not materialized. Recently, another area has shown much more pragmatic potential: spending classification.
The ability to classify messy, incomplete spending data from numerous source systems into a standardized commodity classification schema (particular for indirect spending categories like services) is critical to perform the spending analysis needed to highlight high-payback opportunities. Historically, the approach has been to go through Accounts Payable (A/P) data and associate a primary commodity code to a supplier, using a vendor like Dun & Bradstreet. The other approach has been to use domain-specific, rule-based classifier systems to directly classify items, inventory, and spending. Two notable vendors here are ePlus and MRO Software. Yet, a generalized solution to provide the proverbial black box for spending classification has been elusive.
Recently though, vendors like SoftFace, Zycus, and Ketera have been working to solve this problem. SoftFace and Zycus come from the content management space and are furthest along. SoftFace has performed a few implementations, and Zycus just announced its implementation at General Electric. AMR Research talked to a few of these vendors’ clients, and the technology does appear to work pretty well, classifying spending data into a United Nations Standard Product and Service Code (UN/SPSC) taxonomy at an 80% to 85% accuracy rate. Since both vendors are taking ePlus’s proven approach for aggregating knowledge gained from implementations back into the standard product, these rates should improve even more with time. The products are still early and need improved integration, user-friendliness, and analytics; however, these are not showstoppers.
Zycus uses a statistically based inference engine while SoftFace uses a sort of turbo charged rule-based and pattern-matching combination--used in conjunction with generalized and specialized dictionaries. While both vendors will argue about which has the best black box, both seem to work in the field. Given the high impact of this technology, these vendors have great opportunities selling not just to end-users, but to services providers--like sourcing consultants--who can spend less time collecting data and more time analyzing it and helping users realize the savings.
If a user sees the need to auto-classify into a standard taxonomy like UN/SPSC, the main consideration (beyond collecting the raw data) is whether this should be a one-time or periodic service, or whether it should be bought as an analytic engine. SoftFace has a more product-centric approach while Zycus (based out of India) seems more willing to do a services-oriented project. Right now, the market favors a lower-cost, one-time service approach. However, over time, a product focus will take precedence, especially in the areas of e-procurement and contract management (contracts need to get similarly digitized and have their content classified).
In two years, an e-procurement best practice will be the real-time auto-classification of spending at the time of requisition. Rather than constantly reading through purchase history to chase users who misclassify spending in the first place, the future lies in the ability for an Enterprise Resource Planning (ERP) or e-procurement vendor to present an end-user with real-time auto-classification during requisitioning. In this way, the flexible workflow capabilities can then be leveraged to guide the purchase to the right contract or commodity manager. ERP and e-procurement vendors need to look at this as a real-time Web service to bring to their users, and thus find a way to integrate their requisitioning functionality to these new auto-classification vendors.
Zycus is a leader in Spend Data Management™ (SDM) with a compelling vision for enterprises who plan to transform their procurement organization and stay at the cutting edge of Supply Management. SDM is the foundation to cleanse, classify and enrich vendor, master and transactional data that powers on-going global spend analysis. Fortune 500 organizations such as General Electric, Dupont, 3M, Alcoa, US Federal Government Marketplace (GSA) and Unilever along with more than 35 Fortune 500 organizations have already implemented and experienced the power of Zycus software and solutions.
For media enquiries, contact:
Anurag Dixit
Director – Marketing, Spend Analysis Solutions
Zycus Inc.
Ph. +1 888 451 0145;
Email: anurag@zycus.com
|
Back
