In the first two parts of this mini-series (Part 1 and Part 2) looking at Oracle's new spend classification tool, I examined some of the basic capabilities it presents as well as how it appears to work in practice based on a relatively quick demonstration. Today, I'll continue my analysis looking at how it compares to other solutions in the market. As I said at the start of my analysis examining Oracle's Spend Classification capabilities based on my quick look at the tool, I'd say on a first pass, it compares quite favorably with other classification approaches in the market. But it's important to step back for a minute before drawing direct comparisons. Why? I believe it's important to call out that there's a philosophical distinction between classification approaches and organizations must decide which camp they’re in from the start of the process before deciding which vendors to short-list.
Many applications focus first on enabling a solution-driven approach to classification. This often combines a tool with offshore resources for manual data remediation to increase classification accuracy and coverage. But Oracle, along with Spend Radar, BIQ and some other tools, are entirely user-driven and do not require additional services (unless an organization wants to engage a services provider to manage the classification process). Granted, it's possible to work with providers like Ariba, BravoSolution, Emptoris, SAP and Zycus and just license the software (or aspects of the software, depending on the provider), but their overall philosophical approaches to classification and broader spend visibility enablement are often set up around a solution-driven classification model (at least on the initial pass of the data, and quite often, for subsequent refreshes). Not that any approach is any better or worse -- every company's own needs and experience should determine whether or not they want to take ownership of the data classification process themselves or take a more outsourced approach.
I believe the more important question here is whether or not companies really can -- or should -- care about buying spend classification as a stand-alone capability. For many, it's clear that the greatest need is a broader spend visibility solution that goes beyond just extraction, cleansing and classification to encompass broader enrichment (e.g., diversity, risk, sustainability), analytics and reporting requirements -- and potentially more. Still, some companies that want to leverage existing BI investments or use an open source BI reporting tool -- come on, admit it, you know who you are -- on the front-end might just want the initial classification component. Still, at least Oracle can now serve both segments of the market (which was a significant gap, previously).
When it comes to Oracle's classification engine, how does it stack up to the rest? Short of a true bench test, running multiple classification engines on the same spend data set, it's impossible to say. I will say, however, it definitely provides more of a closed, black-box experience than some of the other approaches in the market which are more hands-on and intensive (requiring greater time and effort -- you get out what you put in). At the end of the day, there's not a "right way" here. And ultimately, when it comes to classifying data at the line item level, the actual approach (AI, rules-based, Bayesian learning, etc.) is less important than the ability of the tool to get the job done quickly, and most important, accurately.
If Oracle can build a strong reference list in the coming quarters, I suspect they'll have a winner on their hands. At this stage, I would recommend that all Oracle procurement customers add Oracle to their short-list of potential spend classification providers. Other companies might be curious as well to see what Oracle has, but until the solution is fully referenceable, it's always worth being a bit cautious, and I'd leave it to the installed base to be the guinea pigs first. Moreover, it will be interesting to see how Oracle customers that have previously used perennial spend classification providers like Zycus respond to Oracle's approach, especially given the fact it could prove -- at $40K per seat -- quite cost effective if companies limit the use of the tool just to key individuals. Perhaps this will end up driving additional price pressure in the market, particularly in the case of larger spend sets that providers have traditionally bid on, at least in part, on overall size.