The MLA Loop

Once you have chosen topics of interest, one possible approach to pursuing an outcome goal is to adopt a Measure Learn Act (MLA) cycle.
Image without caption

M = Measure

Data are very often in a raw state. Reading them one by one does not help. Seeing them in a graph in dashboard can help, but it is still not enough to make an assessment.
You have to filter them, process them, aggregate them, and in the end you get a number, a measure, that expresses what you want to achieve.
Do you want to improve the health of your product? Start by calculating a number that expresses how properly the product is performing + compliance with the designated maintenance activities.
Do you want to improve the energy efficiency of your product? Start by calculating a number that expresses consumption in relation to the required output.
The key is to have one single measure and to calculate it on a regular basis, by processing the data stream.

L = Learn

You learn in 2 steps
  1. Work with the measure you got. Observe its trend over time. Compare the value with the average of the previous X days or the entire life of the product. Compare the value for one product with the average for all products of the same model. Segment and compare that value by customer, by service center, by geographic region, by any other tag that make sense for you.
This will help you answer questions such as:
  • are we generally getting better or worse?
  • which products are getting worse/improving ?
  • which customers are getting worse/improving ?
  • in which areas or for which teams are the products getting better / worse ?
  1. Work with all the raw data that led to that measure. This is when dashboards help, to analyze the raw data, to find the details that help you understand the "why".

A = Act

You can act at 2 levels:
  1. Based on what you have learned, the whys you have discovered, you can act accordingly, based on your knowledge.
  1. Based on your prior knowledge and what you learned in step 2 you can instruct an expert system that automatically suggests the action to be taken. This is a fundamental step to distribute actions to any user who doesn't really know what to do, even if they look into data
In this way you have just kicked off the loop.
Now keep measuring. Regularly check to see if the measure improves. Learn new "whys" based on new situations or new products. Adjust your actions and expert system accordingly. And keep the loop going to reach your target. Once you have reach it, keep it going again to maintain it in the long run.