Daily Archives: April 26, 2008

Spend Analysis Meme Busting, Part I

Today’s post is courtesy of Eric Strovink of BIQ. (It is also a good summary of some of the more critical assumptions that were wrong in the book I reviewed last Friday in this post.) 

The following memes are circulating through the space.  Needless to say, they are all very false!

1. Vendor classification by spend is the secret sauce.
For direct materials, the key is said to be automatic UNSPSC classification. For indirect materials, the key is said to be D&B-style “who owns whom” plus SIC code. This ignores the fact that who owns whom is often irrelevant (Hilton hotels, for example, are all franchise-owned), and that many large vendors sell multiple, disparate products and services, so vendor mapping by itself is often ineffective.

2. Direct linkage to existing accounting systems or e-commerce trading platforms is the “holy grail”.
The logic is that the transactions from the e-commerce platform or ERP system should feed the spend analysis system transparently and directly. The confusion here is between “business intelligence” and “spend analysis,” a distinction that is often incorrectly blurred. To do its job properly, a spend analysis system must re-cast and re-map transactions into a useful form, often in a very different manner for each individual analysis, not just report blindly on fixed input data. A direct linkage to accounting is therefore a bad idea.

3. Spend analysis means the propagation of spend data to a large audience via a data warehouse.
This is the idea that a shared data warehouse is a “spend analysis” system, when in fact no useful analysis can be done without alteration of the schema (something that is impossible in the data warehouse model).

4. Conventional relational reporting tools are useful on spend datasets.
In fact, reporting on spend data can’t be relational, because only OLAP queries will function reasonably on large spend data sets. And, conventional reporting isn’t useful, because useful analyses always involve modeling and data outside the scope of the spend analysis system.

5. Spend analysis is only useful within the context of an e-sourcing suite.
This is a fallacy promoted by suite vendors. Armed with good visibility into spend, any procurement department can improve performance dramatically without a multi-million dollar commitment to an e-sourcing suite. Furthermore, the actual proportion of spend actually traveling through the suite is typically less than 20% of total spend. Thus, “integration” with a minority of the spend is a step backward, not a step forward.

6. “Point” or “best of breed” solutions should be avoided; rather, they should be part of an ERP system or e-sourcing suite. This might be true if the point solution is expensive or complex to implement. It is not true if the solution is inexpensive and easy to use. This is especially true given that suite vendors promote warehouse solutions that offer little or no analytical capability. In that case, the point solution can add significant capability that is lacking.

7. Vendor familying and spend mapping is difficult and/or requires special technology and tools.
This is a marketing myth, used by spend analysis vendors to aggrandize their own tools and/or expertise. Procurement professionals and sourcing consultants are far better at classifying spend than the “10 guys in Bangalore” behind most vendors’ claims. The familying/mapping process can be accomplished easily and quickly in most cases, even on large datasets.

8. It’s critically important to get spend mapping “right” the first time.
Wrong. No spend map survives first contact with a procurement professional who knows his commodities well. The best strategy is to empower that procurement professional to correct errors as he finds them, quickly, easily, and with real-time results. Nothing else will satisfy him.

In summary:

1. The “secret sauce” is being able to build the cubes you need.

2. There is no “holy grail” of integration.

3. Spend Analysis and Data Warehouse technologies are two completely different things.

4. Spend Analysis and Spend Reporting are two completely different things.

5. Spend Analysis can be done on its own and does not need to be part of a sourcing suite. 

6. Best-of-Breed solutions are just fine for spend analysis.  

7. Spend Mapping is easy.  The real secret sauce is “map the GL-codes, map the vendors, map the GL-code and vendor combinations for vendors who supply more than one GL-code”.  This gets you at least 90% and can be done by an accounts payable clerk in a day in most organizations with a proper solution that supports rules-based mapping and rule (group) priorities.

8. No one gets spend mapping “right” the first time, so there should be no assumption of such.  That’s why a real spend analysis solution where you can continue to build cubes and throw them away until you’re happy is critical.