The Delphi Can Finally See Spend -- Oracle Unveils New Spend Classification Capability
Earlier today, Oracle announced a new solution that could not have come a decade -- my slip, I mean day -- too soon. Oracle's new spend classification tool, which the provider is bundling in with its spend analytics solution, will allow companies to classify spend from multiple information sources, included but not limited to Oracle source systems. Perhaps Oracle finally felt the need to bring something to market given SAP's Analytics, Inc. acquisition earlier in the year.
Oracle's new classification offering marks a complete 180 degree turn from the days when Oracle argued that companies should punt on the spend analysis issue entirely or embrace a data warehousing approach (yes, if you can believe it, this was once a policy stance). Moreover, Oracle is the last of the major Spend Management systems providers to offer a spend classification capability as part of its solutions -- Zycus, Ariba, Emptoris, SAP, BravoSolution and others have offered similar capabilities for quite sometime. Most recently, Spend Radar entered this market as well, and has shot out of the start-up gate at breakneck speed (moreover, it's also possible to classify spend in systems such as BIQ if you're somewhat spend adept).
But how exactly is Oracle putting a stake in the ground with its new offering? If you want it straight from the Delphi's politburo, here's the propaganda verbatim as released on the wire earlier today: "A companion to Oracle Procurement and Spend Analytics, Oracle Spend Classification is a new module of the Oracle BI Applications that helps procurement departments categorize spend into a target taxonomy. Featuring a knowledge base that analyzes an organization's buying patterns, Oracle Spend Classification uses Oracle Data Mining for machine learning and other predictive techniques to automatically categorize spend. Through improved spend classification, sourcing managers gain a clearer picture of spending across the organization and are better able to identify cost savings opportunities as well as increase negotiation leverage with suppliers."
While I've not had a chance to look at the tool yet -- I hope to at some point in the next week -- I'm guessing it's nothing revolutionary given the fact that there are numerous approaches to spend classification in the market that already work quite well. After all, spend classification is not rocket science (even though I'll agree it's possible to use intimidating phrases such as "machine learning" -- or was that AI or rules-based -- to obfuscate the basics of what classification, regardless of approach, really does). Still, the fact that Oracle's finally got an offering in this area should be music to the ears of Oracle shops looking to have another option besides best of breed providers (not to mention better educating the IT organization on the role of end-to-end spend visibility requirements). In any event, we'll need to save a deeper dive on the solution for a post-demonstration review, which I'm hoping to post next week. Even if I can't get Oracle to go on the record or show the product in a timely fashion, I'll still get what I need from other sources. So don't worry, I'll be sharing the details -- hopefully more good than bad -- soon enough.
- Jason Busch
Tweet
TweetBacks
But back when I had a role at Freemarkets/Ariba, do you know how we did our "machine" spend cleansing? Yeah we shipped it to India, just like everybody else does. We paid a bunch of Indians like .50 per hour to log into the web to figure out what the suppliers did and give them a UNSPSC. Much like I'm doing it today. At a tad bit higher than .50 per hour.
But to pretend like any of this is automated is a complete joke. Even after years of experience with Indian cheap labor we still hadn't figured out how to apply that work to the next dataset of 10,000 transactions.
The ENTIRE POINT of BIQ is to allow you to do complex things by yourself -- such as construct your own datasets, modify them on the fly, build your own analysis models, and so on. Mapping spend is a simple point-and-click operation performed directly from the BIQ Viewer, and anyone can learn to do it in just a few minutes.
There are excellent BIQ services providers who will cheerfully classify spend for you (and they may well be better at it than you are). They can help you with opportunity assessments, or they may use BIQ to perform complex analysis that you may prefer to outsource.
However, there are BIQ end users who prefer to do all of these things by themselves, and there are also folks who use a mix of outsourced and insourced personnel. That's the whole idea -- do as much or as little as you want, but you always have the option to do it yourself.
Thanks for chiming in. It's been my observation that when people tend to gravitate toward BIQ, there is a perceived need to get to a new level of analysis. The ability to classify and normalize dirty spend data does not immediately come to mind, although yes, I know, it’s possible, as part of the process. It’s also been possible to classify datasets using standard MSFT tools for quite sometime provided you put the hours into it or have someone offshore do it (as the first two commentators pointed out).
My main point, though, is that most of the market these days seem to put more stock in what I’ll term the first two of three steps of spend visibility, load and classify. Whether this is hogwash as a stand alone thing I’ll leave up to the market to decide, but I’m continually surprised by how much people think this is their greatest spend challenge – not the analysis component (after all, that's what incumbent uber cube approaches and BI are for, right ;-)
Cest la spend vie. Personally, I say the sooner you can get to "drill baby, drill," the better, but I know such a perspective is not popular at the moment. Except in more advanced companies when it comes to spend and Alaska when it comes to you know what ...
Yours on the rig, Jason