In God we Trust, for Everything Else There’s Data

Statistics are all about data. But did you know that data can be manipulated to provide you with the results that you thought you should have had in the first place? Don’t get me wrong. I believe data is very important. Without it, we’d be hard pressed to provide evidence in specific situations. However, if data is not collected in a controlled manner, the data can be useless. Let me explain.

To collect data, one needs to first understand the purpose for doing so. And in business settings, the place to start is to identify low performing processes. Collecting data on these processes will provide you with the evidence you need to identify areas in the process that need improving. Here are some basic steps to help you with your data collection plan. 

  1. Measure. Identify what you will measure. Is it speed of order entry, quality of service, number of customer complaints, time to process help desk alerts, etc.? Be specific.
  2. Type of Measure. This relates to what you are measuring and can also tell you when you have enough data. For instance, if you’re measuring quality of service, you will possibly measure one or two inputs to the service plus two or three outputs of the service as well as one process measure. The key is to focus on the vital few versus the useful many. Usually, only two or three items will account for over 80% of what is important to the customer; and that’s what needs to be measured.
  3. Type of data. This relates to either discrete or continuous data. Generally, continuous data is the preferred type of data to collect because it gives you information about magnitude. Examples of continuous data are time and temperature.
  4. Operational definition. This is important to determine exactly when data collection starts and stops. If measuring time to complete order entry, when do you start your timer and when do you stop? Is it when the order is received by the order entry clerk or is it when the order is made by the customer? You can see how being precise in this case will enable you to collect the correct data.

There are other considerations in a data collection plan, but the above four considerations are key to ensuring that the data you collect serves the purposes for which it is intended. Other considerations when collecting data for process improvement purposes includes identifying specifications (i.e., the least acceptable measure in the process such as 30 minutes is the upper limit for order entry) and targets (i.e., what customers would consider ideal – this is not always achievable, but it’s good to have stretch goals).

When collecting data, ensure that you have good measurements. This means that data should be easy to understand, they are important to the customer (either directly or indirectly), and the data motivates people to action.