Models are mental constructs we humans use to help us understand and deal with complex realities. In creating those models, we identify what we believe to be the key elements in those realities and the relationships among those elements. Valid models – models that accurately reflect underlying realities – are critical to quality decision making because our decisions are based on those models of reality, not on reality itself.
Populating a model with data provides decision makers with the information they require to support the decisions they must make. A valid model populated with extremely accurate data is ideal. But when a model does not reflect reality, the accuracy of the data that populates that model doesn’t matter. It will generate a false picture of that reality. On the other hand, if the model does reflect reality, reasonable estimates of the required data will result in a relatively accurate picture of that reality.
For example, a person wants to know the area within a circle. He knows that the data required to perform that calculation are the value of pi and the radius of the circle. If he makes a precise measure of the circle’s radius using a laser distance finder, values pi at 3.141592653589793238462643 and enters them into the formula 2πr he is entering extremely accurate data into an invalid model and will arrive at an answer that’s totally wrong. On the other hand, if he makes a reasonable “guesstimate” of the circle’s radius, values pi at 3.14 and enters them into the formula πr2 his answer will be reasonably accurate. So which is more important, the data or the model?
This holds true with the revenue, cost and investment models that are the cornerstone of profitability analytics. If they are valid, they will provide accurate and relevant information on which an organization case base decisions. If they are invalid, the information will be wrong regardless of the accuracy of the data that populates them. For example, there are many manufacturers that have ERP systems that collect and enter extremely detailed and accurate data into traditional, direct labor-based costing models that result in cost information that is not only inaccurate, but also incomplete and often irrelevant. Why? Because direct labor-based cost models do not reflect the economic realities that underlie most manufacturing businesses. (You can read more about the failings of direct labor-based cost models at http://formingworld.com/labor-based-costing/)
A model that represents the fundamental economic realities that underlie an organization’s operations will result in reasonably accurate information even if populated with estimates. The same can’t be said for an invalid model populated with perfect data. Both models and data are important. You can’t just throw darts at a dartboard to develop your data. But models are more important to decision making than the data that populates them.
Do you agree?