This series discusses the recent report from CAPS Research on the role of optimization in strategic sourcing. The primary goal is to highlight, clarify, and, in some cases, correct parts of the report that are important, confusing, or incorrect to insure that you have the best introduction to strategic sourcing decision optimization that one can have.
This chapter starts off by explaining the buyer and supplier data requirements for decision optimization and it does a good job. The five data requirements it lists for a buyer are spot on:
- accurate historical data and projected volumes
this allows you to not only create accurate baselines, but to perform a “sanity” check on the demand forecasts you are given
- complete list of requirements
these will form the foundations of your model constraints
- minimum quantities and timeframes
these not only specify minimum model awards, but help you determine what suppliers are qualified
- complete specifications
these are necessary for the suppliers to submit accurate bids
- identification of locations and individual demands
these define the minimum set of bids your suppliers need to submit
Next it goes on to discuss a number of optimization model issues including those of model size, complex sourcing events, small and standard buys, and the optimization sweet spot. The report did a good job on these issues, but three points need to be clarified.
While it is true that many moderately sized problems will still be challenging for a desktop or laptop, moderately sized problems can easily be solved in a matter of minutes (and sometimes even seconds) on average mid-end servers. Furthermore, today’s high-end servers can handle problems that are quite large indeed. And while it may still be the case that no single provider can handle all of your strategic sourcing decision optimization needs, the 80% solution is still a great one — license a solution that gives you 80% coverage and then utilize a second provider for large, custom, high-dollar events where the ROI will dwarf the additional cost.
The report correctly states that while it is quite possible that the software will not find ‘provable’ optimal solutions for the model, the software can nearly always find good solutions that will be ‘near optimal’. However, it does not state that those ‘near optimal’ solutions will be ‘provably’ near optimal, which is always the case with MILP optimization (that is the foundation of the majority of strategic sourcing decision optimization products on the market today). Since MILP solvers start by finding the optimal solution to the relaxed linear model, the distance of each successive solution from the absolute lower bound (which can be increased every time a solution sub-space is fully explored) is always known. So even though there may not be enough time to fully explore the potential solution space and find the provably optimal solution, the solution returned is provably near optimal within a certain tolerance.
Finally, the statement that most e-purchasing suites have an optimization module that can address the large number of bidding opportunities in this area is laughable. There are dozens (and dozens) of providers who offer e-sourcing and / or e-procurement suites (just check the resource site), but only a handful that offer (true) strategic sourcing decision optimization. When it comes to strategic sourcing decision optimization, you’re pretty much limited to Algorhythm, Bravo Solution (Vertical Net), CombineNet, Emptoris, Iasta, or Trade Extensions … only four of these can be considered suites … and only four are self-service. While I expect that we will see more providers with true optimization offerings as part of their suites in the future, until the utilization of strategic sourcing decision optimization becomes mainstream, I don’t expect any new providers will emerge.
Next Part IV: Implementation Issues
Share This on Linked In