Spend Analysis I: The Value Curve
Today I'd like to welcome Eric Strovink of BIQ who, as I indicated in my There's No Spend Analysis Without the Slice 'N' Dice post, is going to be authoring the first part of this series examining what is required for a true spend analysis system, spend analysis 2.0 if you are part of the 2.0 movement, as opposed to just a basic spend visibility system.
Spend Analysis has always suffered from what the late British humorist Stephen Potter might have called the "So What Diathesis." In other words, now that you have your spending loaded and classified, what next? Well, if you've never seen your purchasing data loaded into a spend analysis system, you're in for a treat, because you can find savings opportunities just by drilling around. It's often that easy -- drill around; find opportunities.
However, once the low-hanging fruit is harvested, which can take anywhere from 6 to 12 months, the value of the spend analysis system declines steeply -- at which point Mr. Potter's observation comes home to roost. As illustrated below, there is a moment at which the cost of the spend analysis system begins to exceed its ongoing value.
It is shortly after this time that (1) usage of the product drops to low levels; (2) the rest of the organization begins to question the value of the software; and (3) stakeholders come under pressure to justify continued high expenditures.
That's why it's odd to hear people talk about "The Spending Cube" -- in capital letters -- as though there were only one data cube ever to be built. Actually, there are many different ways to look at spend, and there's lots of spend data that simply can't be organized into a single data cube anyway. How about a compliance cube, oriented around invoice level data? A purchasing card cube, specific to p-card idiosyncrasies? A T&E cube, built from travel agency data on "best price" versus "actual price," tracking employee travel and the reasons for the discrepancies?
In fact, it's obvious to anyone who has worked with multiple datasets at the A/P, PO, and invoice level that there are many, many different kinds of data to analyze. Each dataset addresses more opportunity, and presents another chance to apply a sophisticated analysis tool. Some of these datasets aren't "spending" datasets at all, but consist of demand-side information -- for example, cell phone or fleet vehicle usage records, or operational data such as equipment recovery and maintenance logs.
If a spend analysis system makes it easy to load data and create new datasets, which it should; and if the system supports as many datasets as you'd like, as it ought; then there really isn't any limit to how often the system can be used, or to how many different kinds of data it can be applied. Which means that a full-utilization spend analysis system value curve looks more like this:
In other words, each use of the spend analysis system provides high initial value, as well as residual value; but the system is used again and again for new sets of data. The value of the spend analysis software therefore remains high over time.
Next installment: The Psychology of Spend Analysis