By now, what I once coined “Corporate America’s dirty little secret” is a secret no more. Poor visibility is as pervasive as Janet Jackson; and it’s wreaking havoc on supply management performance everywhere from L.A. to Tokyo.

Lack of visibility into timely, detailed, and accurate spend data limit spend leverage, hinder compliance, and cause supply managers to develop sourcing and supplier strategies in the dark. Throw in tightening supply market dynamics and tougher scrutiny from CFOs and regulators, and the spend deficiency syndrome is now more exposed than Miss Jackson at halftime.

Considering these factors, it’s no surprise that the nearly 300 supply management and business executives attending Empower 2006 listed improving visibility into timely and accurate spend data as their top initiative for the next two years.  

I won’t get into a long-winded diatribe on the hurdles and best practices for spending analysis. I’ll leave that to the professional prognosticators. In fact, Aberdeen Group’s most recent study on the subject – The Spend Intelligence Benchmark Report: Turning Data into Action – does a smash-up job of dissecting the issue.

Instead, I’d like to share a true story of one company’s journey to spend intelligence. For the purposes of this story, the company, whose name will be protected to avoid divulging their competitive edge, will be referred to as Global Tech Co.

Once upon a time back in the last century, Global Tech Co. realized it had a problem. It had poor visibility into how much it was spending, with which suppliers, and on what products and services. Spend data was stored in multiple, disparate systems across the enterprise. Collecting and analyzing this data was very labor intensive, so the company’s supply management team defaulted to doing periodic spend analysis projects that only examined a fraction of its spend data.

Global Tech Co.’s IT leadership assured the supply management group that the spend visibility issue would be resolved when the company standardized on a common ERP system. After the fact, the supply team quickly realized that ERP systems are built for high-level financial analysis and lacked the detailed, line-item attribute data needed for meaningful spend analysis. For example, ERP standardization reaffirmed what the supply team already knew – Global Tech Co. spent a lot on hardware. But the ERP data didn’t help commodity managers determine how much was being spent on high-end SPARC stations versus laptops. Nor could it shed insight into whether these items were being bought for development engineers or office administrators. (Earlier this year, SpendMatters’ Jason Busch posted on the follies of the ERP-only approach to spend analysis.)

Global Tech’s next strategy was to assign standard part numbers to the items it purchased.  This led to a proliferation of part numbers as well as differences within the part numbers. To continue our example in the area of hardware, part numbers were difficult to assign due to different models and configurations.

Global Tech’s next move was to dump spend data into a data warehouse with a snazzy reporting tool on top. This allowed the supply team to slice and dice spend data and create reports to impress their boss and friends. However, this approach still didn’t solve the core symptom: Global Tech lacked spend data detail that could support accurate classification and meaningful analysis. (Garbage in, garbage out, anyone?)

As a result, the supply team spent their time pulling spend data from the warehouse into spreadsheet pivot tables so they could sort it by attributes. But with key detail – like processor speed or memory – buried in the description, this information was hard to dig up. This lengthy and labor-intensive process forced Global Tech to continue to conduct spend analyses on a project basis. Worse yet, time and resource constraints limited analyses to Global Tech’s big-ticket spend categories.

Finally, Global Tech Co. adopted a solution that automated spend data extraction, classification, and analysis. Through the use of business-rule and category-libraries, the tool rapidly accessed and evaluated the line-item spend attributes and classify these to a common schema.

[Because it mostly bought common indirect goods and services, Global Tech Co. elected to map spend data to the industry standard UNSPSC. However, UNSPSC has proven too limited for use by many companies, particularly manufacturers. We will analyze the pros and cons of UNSPSC in a future post.]

This automated attribute-based classification gave Global Tech Co. the detail required for meaningful analysis, enabling the supply management team to improve spend leverage and develop informed sourcing and supplier management strategies. For example, the company has been able to change its hardware buying policies and practices, assigning approved hardware types and configurations for each role in the organization. (No more SPARC stations for the receptionist!) It also uncovered more spending power, such as hardware purchases that had been misclassified under office products.

Such insight and detail enabled Global Tech Co. to achieve additional 5% - 15% cost savings – depending on the spend category – through improved spend leverage, supplier rationalization, and new buying rules.
This automated spend intelligence approach enabled Global Tech Co. to be more systematic about spending analysis. The company auto-refreshes 100% of its spend data pool on a monthly basis, arming its supply team with the intelligence required to make and execute informed sourcing and supplier strategies across a much broader portion of Global Tech’s spend.

The moral of this story (in case you missed it) is not about cost savings. It’s about understanding that the success of any spend analysis initiative has less to do with reporting and analytics (although chief execs always like a story with pretty pictures) and more to do with having the proper tools to extract, cleanse, classify, and enrich spend data at a detailed level – all on a repeatable basis.

Building your spend analysis initiative without a repeatable process for detailed data classification is like building a sand castle on the shore. It may look nice momentarily, but it will certain erode with time. And it will take heck of a lot of effort to rebuild.