Making sense of segmentation
Now that our suppliers are segmented into groups we need to determine the interventions needed for each group. However, in practice, there are degrees of importance and the demarcation lines between one group and another are more blurred (Figure 4.7). Moreover, intervention can take many forms; one size does not fit all so segmentation not only determines who is important, but the outputs from segmentation, and specifically the basis upon which a supplier has been deemed important, determine the degree and nature of intervention required. As we develop approaches for a specific supplier we therefore need to refer back to the basis upon which they were determined to be important. For example, if a supplier is important because there is risk to our brand due to practice in the supply chain then our interventions need to focus on the supply chain (eg through SCM); a supplier who is important due to current knowledge of our business and a high spend might need close management (eg through SM).
FIGURE 4.7 Degrees of importance and blurred demarcations
When sat is a room, at the end of a long team-based segmentation workshop, surrounded by flip charts of notes, scores and lists, it is easy to wonder what to do next. It is in fact the discussion and debate amongst the assembled team during the workshop that is the most valuable - the outputs serve as a record of this discussion. Possibly the greatest challenge within segmentation is making sense of these outputs so that the right interventions for specific suppliers can be developed.
The problem is that the segmentation process produces five separate, independent scores for each supplier. Summing these to produce a grand score isn't useful as we could risk excluding a supplier who has a low score overall but yet scores highly against one key criterion such as risk, and therefore requires intervention. Instead our segmentation process must preserve the scores against each criterion for each supplier but we need to be able to assimilate all these scores, across multiple suppliers, in such a way as to prioritize who we should spend time with.
It is possible that a complex mathematical model could do this, however I don't believe I have ever seen one that is completely effective. Segmentation in fact requires good old brainpower and judgement informed by a something that allows us to assimilate and process vast amounts of information. This is not a unique problem; marketeers face similar issues when attempting to analyse a competitor landscape or compare product attributes, so they use visual tools. Typically where complex variables and information needs to be assimilated visual tools tend to be most effective at providing a basis for human judgement. Visual representations of the individual suppliers' evaluations against the criteria help to allow rapid multi-supplier evaluation, ideally performed in a carefully chosen group, to determine who is important and why. This is simple to do and if we take our segmentation score charts and mark up the scores on each, then join these together it creates a unique shape. Figure 4.8 show two suppliers with two very different shapes created during segmentation. On the left, an outsourced partner with high spend who is important today and for the future; on the right, a supplier who presents a significant degree of risk to us and where there is high spend. So which one should we spend time with? Well actually both, but in different ways: for one we need to build a relationship and the other we need to assess and manage risk. The point here is that using a visual method allows us to lay out all the suppliers in front of us and compare one to another to decide who needs what intervention.
There are many other types of graphical representation that could be used here with the segmentation scores and Figure 4.9 shows some examples. A good old bar chart or radar diagram are sound and easily accessible using Excel or similar; however, I favour the 'coxcomb graphic', apparently first used by Florence Nightingale to illustrate levels of mortality over time, but increasingly popular with statisticians and communication experts looking for a good 'infographic'.