In my last post How to Begin an Analytics Transformation in Your Organization - Part 1, we reviewed a visual that I’ve used to describe the different stages of an organization’s analytics transformation. I also went into some detail about the advantages of limiting your scope to a receptive team that already has analytics in place and would benefit from automation. In this post, I will delve into what types of analytics define the different stages of a company’s evolution and go into some detail about the first, and most critical, stage.
I use the word evolution - rather than enhancement - because the types of actions that take place during this stage result in both added (or improved) functionality as well as changes to processes. During those modifications to systems and procedures, I try to be a force that encourages my users to see analytics differently, with the
intended result being lasting cultural change. Such things don’t happen overnight and, in fact, should be occurring continuously. Nothing is ever perfect, especially culture, so habitually analyzing and revising existing paradigms at an individual, team and organizational level is an exercise that can regularly provide valuable insights.
In general, I see analytics evolving through four broad stages defined by the type of analytics being used and aptly compared to the development of a human being:
1. Descriptive – newborn infant
2. Diagnostic – curious toddler
3. Predictive – turbulent teen
4. Prescriptive – competent adult
Descriptive analytics will tell you what happened in the past. This is the simplest form and, in my experience, is better understood using the term reporting. When an organization can effectively produce accurate descriptive reports, users should have the ability to aggregate their relevant metrics and slice them by whatever unit of analysis or timespan they require. A good example would be a manufacturing company that can tell you how many widgets were assembled during any specific day, week, month, or year and provide breakdowns by assembly line, location, shift, or any other relevant grouping. The chart below is a good example of this type of reporting.
This is the first stage of any analytics transformation and, as such, it is often loaded with some laborious tasks that will be leveraged in later iterations. Such tasks include the implementation of a continuous data validation framework, working with your security team to discuss access policies and consider privacy constraints, and developing a user-friendly data dictionary and data lineage documents. If your users are all speaking the same language, know where their data is being sourced, and can trust that it’s accurate, you’ve already overcome three major hurdles in the development of organizational analytics. Engaging your security team early and often mitigates the risks associated with unauthorized access to data and, in many cases, will be a necessary step to ensuring compliance with regulatory or audit requirements. However, it is critical that you discuss these processes with your executive sponsors. Much of what I just mentioned falls within the scope of data governance and the last thing that you want to do is duplicate work or undermine existing policies.
This too is when your client is new to analytics and, hopefully, open to considering new ways of operating. Use this period to build the proper framework for evolving. Most investments have an expected return so find ways to measure the results of those efforts and enable continuous improvements. This can be an exciting time for any organization and when well executed, that excitement can spread like wildfire. The best architects I have known can identify when this is happening and are able to capitalize on it to increase engagement, investment, and adoption.
Please stay tuned for Part 3 of the How to Begin an Analytics Transformation in Your Organization series during which I will discuss diagnostic analytics.
Comments