Aberdeen’s report on Spend Analysis (“Spend Analysis: Working Too Hard for the Money,” available free from Iasta and others) draws useful conclusions about the forest, but then, like many other studies in this space, loses its way in the trees. Consider this reasoning:

• Wealthy people tend to be successful.
• Wealthy people typically drive luxury automobiles.
• Therefore, wealthy people are successful because they drive luxury automobiles.

Aberdeen seems to reason about spend analysis in the same way:

• Best-in-class purchasing organizations tend to have bought spend analysis systems.
• Best-in-class organizations typically have bought spend analysis from one of the “Big 3” SA vendors.
• Therefore, organizations are best-in-class because they bought spend analysis from the “Big 3.”

Thus, when the report uses survey data to uncover the underlying reasons why best-in-class organizations are successful, it loses its way. Using circular reasoning, it offers up precisely what one would expect: the standard “Big 3” marketing messages. Those messages, most of which haven’t changed in years, are these:

1. Automated spend classification. “Only we can classify your spend, with automated algorithms and Bayesian analysis and special databases and… and… well, the point is, you can’t do it yourself, you have to hire us.”
2. Standard reports. “Our suite of standard reports is better than anyone’s. Why, you don’t even need a sourcing consultant or a sourcing expert on staff, our reports tell you exactly what to do.”
3. Integration with RFx. “Buy our suite, because it’s all ‘integrated.’ Just ignore the fact that spend analysis doesn’t ‘integrate’ with RFx, that’s not important now.”
4. Integration with Contract Management for ‘compliance.’ “Let’s fail to mention that you can create a rules-based Contracts dimension yourself in just a few hours, whether you have a CM system or not.”

I’ve addressed these points at length elsewhere (see, for example, Common Sense Cleansing and What Purchasing.com Got Wrong), so I won’t do it here, except to point out that every SA application worth its salt creates a rules system that automatically classifies and maps spend. That’s the whole point, after all. Aberdeen’s report confuses the process of rules system creation with the results of rules system creation. Once a rules system is built, by whatever methods, the end result is a system that automatically maps and classifies both current and future spending.

The above nothwithstanding, the report contains a fascinating chart that shows survey respondents’ opinions of the “importance” of data analysis, data management, reporting, and supplier content, plotted against those same respondents’ classifications of their “current ability” in those four areas. It appears that “current ability” deeply lags “importance” in all four of them. As Aberdeen says,

While organizations recognize the advantages that can be gained from technology deployment for spend analysis, they have still not bridged the gap between theory and practice. Across the primary steps in the spend analysis process, enterprises are generally unable to fully leverage their spend analysis solutions… [emphasis added]

Aberdeen fails to draw the obvious conclusion from this — namely, that legacy approaches to spend analysis are disappointing their users across the board, despite causing an uptick in procurement efficiency. This result ought to be a key conclusion of the study. Spend analysis is about analysis, after all, not about the mechanics of data preparation. In fact, the four key components of spend analysis are:

• Powerful analysis and ad hoc reporting tools (“data analysis” and “reporting”)
• Flexible and ultra-fast dataset creation (“data management” and “supplier content”)
• Real-time dataset modification (“data management,” “data analysis,” and “reporting”)
• Flexible deployment (Aberdeen doesn’t address this, but the SA space has changed: powerful spend analysis is now deployable for small dollars, on individual analysts’ desktops, without an organization-wide commitment).

All of these components are interdependent — for example, you can’t perform ad hoc analysis if you can’t quickly change the structure of a dataset. And, you can’t change the structure of a dataset if it’s shared with others, because the other users certainly won’t appreciate you changing things out from under them.

It really should be old news by now: data extraction, transformation, loading, familying, and mapping are processes that are easily automated by in-house personnel using modern tools, or by outsourced resources using those same tools. It’s a shame that Aberdeen chose to focus on the “old think” of cleansing — only the very first step of a spend analysis effort — rather than pursuing the most interesting of its own survey results.