Digital Transformation & Analytics: From Fashion to Reality
Actualizado: abr 19
Which is the best way to go through a digital transformation process by leveraging data knowledge, without dying along the way? It is obvious that by having people with strong data literacy capabilities and a well-chosen technology, companies can reach heaven, but please, Sofi, tell me what issues will I come across with, while swimming in a data lake…in a cold and brave data issues lake, if you can catch the word game.
The first problem is a question: “To be or not to be”. What comes first: Reporting or Analytics? Well, usually reporting comes first: moving from static reports to automated dashboards. Do we need to let our happy times with Excel lie behind and adopt any other innovative tool as soon as possible, just to automate our reports and make our analysts life less boring? After the whole beautiful dashboard is built, is it time to start looking for insights? This is the usual way of working. I’d like to challenge its conception: I think that is more useful, first, to define an analytics framework, build the proper KPIs and visualizations using Py, SQL, R, or whatever language one is familiar with, and then, after an iterative journey of learning, one can build “The Tool”, “The Lab” over a dashboarding tool, in which data will flow systematically, will be visualized in a certain way, and users will experiment, test their hypothesis, and feed great decisions. A tool in which actionable insights will spread along the new formal channels of communication that data allows. Intuitive decisions must be minimized and confirmation bias, definitely killed.
The second issue that arises is what I call: "The short-long term dilemma". How can we leverage a short-term solution, with its own modular improvements along time, fed by the necessary learnings of business questions and data limitations, while we convince companies to keep on investing on data, because they need to develop a long-term robust solution, that will make the most of data and squeeze business opportunities? It is not a "nice to have" asset, it is a toolkit they’ll need; they actually need it now. Well, I think the answer is buried inside the question: as long as the short-term solution is used as a learning device, to gain time and learnings, then the long-term solution will not be built from scratch, then it is not work duplication. As a matter of fact, it will be built over a strong robust base of knowledge, that'll allow actionable insights spread faster, capitalizing their own learning curve. Everything is better tomorrow, but we need to get prepared now, because if we don't, innovation processes of using data to act on market will be a vicious circle.
We will talk in future posts, the key topics related to all these: The development of capabilities to spread across the organization the knowledge data brings, and the encouragement of the usage of the whole suite of features that actual BI technologies provide. From this topic, another question arises: Which technology should be chosen? Which are the drivers: Fashion or Usability? Elegant or Simple Tools? Is cost a key variable to a successful BI process implementation? How can a real applied solution win the battle against fashion?