It’s been seven years since Meta Analytics was published. In that book instead of focusing on the technical aspects I wrote mainly about two main pillars on the insights process.
1) The object: The object is what the subject (the person) will analyze. Understanding the “substance” of what we are analyzing is key to generate a proper interpretation of what is been observe. The object in our case are companies. Companies are systems, a set of things that interact together with a common objective which is making money today (ie Ebitda) and in the future (ie Purchasing intention). The parts that interact together are people, technology, capital, etc. The way in which all those things interact together will be responsible for the level of achievement of the chased objective.
2) The subject: The subject is the person responsible to analyzing and interpreting the data, generate a discovery and tell others that discovery in a way that the person that receives the report understands exactly what it is trying to be told. All these could seems simple, but it’s exactly the point in which everything mess up.
The reality (the observed object) can’t be fully comprehended due to two key variables. I) Imperfect information that prevent us to have all pieces of the “reality puzzle”. II) Mental Models that prevent us to process the information as it is. Once we perceive the reality, it is processed by our mental model which mold that information generating a symbol or representation of the external (Objective) reality.
So basically you have an object. Then you have the subject interpreting that object based on its mental model, and sharing that with other people with different mental models. Quite a challenge right? Well an organization is these multiplied by al the people that is part of it and in a constant change. Information going back and forth being coded and decoded by each particular individual generating what is normally called as “Organizational Culture”.
So, do you get it now? The most effective approach to deal with data is changing your culture, not just hiring more and more data scientists. I mean, is not that having data scientist is not important, because it is. But is not realistic siting a couple of data scientists in a dark corner of your office and wait until they “do their thing”. Managing data is not something that should do the “Data Guys”, we all now have to be data guys. The CEO, the CMO, the CIO, the CFO, the Marketing Manager, the logistics manager, the collection manager, the finance analysts…all of us. The idea of having a BI or Data Science department so you can receive analyze data on the way is not realistic, it didn’t work in the past, is not working now, and it wont work in the future. You have to have access to the data, you have to have some basic knowledge of querying data, to process data and to report data.
But again, this requires changing people’s behavior and changing people’s behavior is way harder than hiring a data scientist, and believe me that I know how hard that is. That’s why just few companies have a data driven culture, so let’s analyze where is the main problem.
Whether you are trying to change your culture by hiring data scientists, or by generating a BI area, what you are doing is putting a few people into an ongoing culture, so basically at some point these people have to decide or adapt their behavior to the company’s normal behavior allowing regular interactions within that “system” or remaining firm in their behavior. That at some point will leave them completely isolated from the rest of the organization.
So, what’s the right move then? The right way, and I believe the only way to facilitate the cultural mutation is to create an Analytics Governance department. Analytics Governance refers to centralizing the generation of findings based on the objectivity of information, within organizational structures, in a specific area that functions outside the organization’s productive structure (technostructure). Setting this area or department outside the productive structure allows the area to:
- Have a vision of the organization as a whole: Unlike the way it works now with the BI department, it’s important that this area can identify the required connections (interactions) among the different parts of the organization, allowing only the information flows that are useful to make each department understands its performance related to the main company objective.
- Cultural change driver (Cultural Backbone): As we said above, if we put a small group of people to interact with the whole culture, at some point the culture will eat them alive and not the way around. The Analytics Governance department, that should be supported and have the main contact with the company’s CEO, can drive cultural changes from the outside, identifying negative behaviors in the companies culture (ie inferences), and generating the right plan to correct that deviation.
- Prevent duplicated work: Since this area have the whole view of the information flows, it’s the right one to identify duplicated efforts on information generation, analysis and reporting, relieving the people or areas operatively working on data (BI, Technology, Data Science, etc).