Working with many companies over the years in a variety of industries, the one constant is that every company has issues with data collection and data validity. First, it can be very difficult to find sources for all of the relevant data needed for supply chain planning. Second, once a source of data is found, it will quickly become obvious that not all of your data can be trusted (e.g. do I really believe that those 25 cases I shipped from Denver to LA only weighed 0.01 lbs?). So how do we create good supply chain plans when our data is imperfect?
My first comment is that we should stop trying for "perfect" data. First of all, it is an unattainable goal. Secondly, even if you had perfect data, it would not necessarily be what you need. Strategic planning is about setting a strategy that will best manage future events. Your data is typically about past events. Your data is useful for strategic planning to the extent that it helps predict future events, but good data in this context is really a means to an end (accurate forecasts), not an end in itself.
With this in mind, my second comment is that sometimes I think we need to have the view of "don't sweat the small stuff". Our ability to predict future events accurately is limited. In the context of strategic planning, where our future forecast may be off by 10, 20, or 30 percent, how much impact does it have if our data is 95, 96, or 97 percent accurate?
- Target the best data sources for information relevant to the question at hand.
- Spend time cleaning the data to remove obvious inaccuracies and outliers.
- Use your data, and other sources, to make you best forecast of future events (demand, costs, capacities, etc.).
- Use reasonable approximations in cases where data does not exist or is no longer applicable.
- Run many scenarios where key parameters are varied in order to test the sensitivity of your answers.
Step 5 is really key. I think many times we can have a false sense of certainty in our answers because we think our data is very accurate. The truth is that no matter how accurate our data is, the future is uncertain. We need to test our assumptions, vary the key parameters and see how our plans change. It is in this iterative process that we will arrive at plans that are likely to serve our supply chains well.