Regarding the business impact of data science related work, as a working ds and team lead a have a few thoughts. There are some sectors and activities that are

Regarding the business impact of data science related work, as a working ds and team lead a have a few thoughts. There are some sectors and activities that are very data science like, to using a blunt anachronism, have been like that for a long time, and have a lot of teachings on how to develop and measure impact on business by the automated decision systems and/or analytical work that they build. Some examples are: - demand forecasting for supply chain processes - credit scoring for loan approval - portfolio risk analysis in financial settings - some misc optimizations work on operations - maybe even six sigma can be listed here Those have always had a data-science-like feel to it. The problem I see is when companies try to implement a data science team is: 1. push out subject expert knowledge and requirements just for the "freedom"; 2. have stake holders to be out of touch with the solutions; 3. too litle focus on putting stuff "in production", tracking, beeing able to experiment, in whatever sense those have for the company; 4. too much focus on numbers that can come out of the ds's computer, and treating operation related numbers as an after thought; 5. no basic knowledge of simple/common/classic solutions for their problem at hand. So yeah, making business impact is way harder than is sounds, and too out of the skill set for the 23yo STEM graduate to actually make impact. And too buzzwordy and impressive for the typical decision maker. I mean, I've heard countless times things like: "if an AI knows how to tell a cat from a dog by using one of those neural nets logarithms, surely it can know how to partition my marketing budget, optimize my coupon giving logic and determine the strategy we should have in order to achieve a very obscurely constructed OKR". (yes, I have worked for people that used the word logarithm to mean algorithm).