Spend Analysis Expectations

Last month, Purchasing Magazine ran a story named Avoid a mess, clean your data which focused on understanding and managing your expectations when launching a spend visibility application. This is a very helpful article for any organization that has designs on moving into this type of technology. The benefits of spend analysis are wide and virtually guaranteed; however, it is a totally different paradigm than eSourcing. Where sourcing is more of a turn-key application, which can be enabled by the vendor within hours and have the company creating projects within days, spend analysis is a completely different animal. Says Ken Hartman of Boston Scientific:

“”The main issue that we’re seeing that I think companies will face is what the industry calls ETL: extract, transform and load,” he says. “The vendors want to show you the analytics and the dashboarding and all that stuff because everybody really likes shiny things, and that’s what they are. What they don’t show you is the greasy stuff that makes the shiny stuff work, and that’s the ETL.”

Ken is a very smart, good guy and straight shooter who I have always enjoyed speaking to. I particularly like his comment because it gives a real testimonial of the heavy lifting that comes with deployment of most spend analysis tools. There is no getting around the fact that the data must be normalized, loaded, cleansed and then cubed. You get the benefits — AFTER all of this is accomplished.

You separate the men from the boys once you have the data. At this point is when your software can parse data on the fly, create new dimensions for viewing, ease the pain for refreshing and merging, and create suppliers links to find the true total spend per item, group or vendor. Most spend applications have advanced to a point that many of the front-end problems are greatly reduced these days and especially SmartAnalytics which streamlines many aspects of the initial process such as translation, loading and chopping of data sources. This drastically reduces the pain of data manipulation.

One Response to Spend Analysis Expectations

  1. Ken’s point is that it needs to be easy to create analysis datasets from the data sources lying around the enterprise. It shouldn’t be a big deal.

    So when a BIQ customer tells me (as one did just yesterday) that they have 24 datasets loaded on their server, and far more floating around the PC’s of individual users — of which only one is an A/P dataset — that tells me the “ETL barrier” isn’t much of an issue any more.

    That’s the way things should be.

Leave a Response