Imagine your job is to put 250 piece puzzles together. You know what the puzzles are supposed to look like when they are finished. You have been putting these puzzles together for a few years so you are pretty good at it. One day you walk into your office to find that you are no longer in the business of putting these 250 piece puzzles together, but rather assembling 5000 piece puzzles that you have no guide for what they are supposed to look like. Oh, and by the way, instead of using your hands to assemble these puzzles you now have to use remote controlled robotic hands.
This is essentially the position that analysts in many sports now find themselves with the advent of motion capture data. Soccer, basketball, baseball and Aussie rules football are just some of the sports that, to some degree, have begun utilizing various technologies that allow them to track everything that moves with a high level of frequency. Instead of utilizing straight play by play or box score data (your 250 piece puzzles), they are now faced with data on the position of everything that moves on the field/court/pitch multiple times a second (the 5000 piece puzzle). Because this data is brand new, nobody really knows exactly what to do with it or what they want to get out of it – thus no defined guide for the data. The data is also incredibly complex and so the old tools that analysts used are often not effective in creating meaningful information out of this new mountain of data, leading the analyst to have to utilize and develop more powerful tools to help create a competitive advantage out of this mountain of data (robotic hands).
The analyst that is placed in this position probably sees the opportunity that they have been handed to significantly increase the impact that they can have on the team. Massive new data sets, when properly translated to information, can deliver a real competitive advantage. What they need though, is a plan on how to make that translation in the most effective way. The analyst has essentially two choices on how to attack the new data.
The first is with analyst driven questions. Analyst driven questions are fun and interesting for the analyst, because they are born out of the mind of the analyst. They often involve a high degree of complexity and, if they were able to describe them to analysts on other teams, would result in a lot of oohs and ahhs from their colleagues. Unfortunately, these are like utilizing the 5000 pieces to put together a picture of the analyst’s dog. The analyst will really enjoy the process and the end product, but it is likely that nobody else will care all that much.
The alternative is to look for strategic questions to answer. Strategic questions are those that are developed with the input of the decision makers at the team. This does not mean that the analyst has no role in developing the questions, but rather that their role is to find out what will be the most useful information that can be culled from the new mountain of data. This is a harder process to get moving as it requires discussion and brainstorming from a variety of people within the organization, and it requires the analyst to push the decision makers in their thinking, so that the analyst can really maximize the potential of the data set, while still delivering information that the decision makers are actually looking for. A process built on this concept will yield information that is novel and utilized, which will lead to the team realizing a true competitive advantage from the new data. This is akin to assembling that 5000 piece puzzle with a communal goal for what it should look like when it is done. Puzzles like that are enjoyed by a much broader array of people than just the person putting it together.
The analyst that is tasked with getting value for their team out of this new mountain of data needs to first remember that their goal should be putting together information that will be utilized and second, that instead of starting by asking questions of the data, they should be asking questions of the decision makers.