Here are a few examples off the top of my head-
One of my first consulting jobs ever was to reverse engineer a bunch of models a former employee of a small engineering company had built. The company’s IT was atrocious, and to work around them, the employee in question had implemented a surprising amount of code in sql queries that were embedded in excel sheets (in general, the ways people work around bad IT make the situation even worse). These models and spreadsheets had become part of nearly every member of her team’s workflow. She left the company, productivity ground to a halt (with no one to keep the spreadsheets up to date, they were suddenly at the mercy of their terrible IT) and they spent a small fortune in consultants to get things back to where they were.
Insurance companies have a lot of data entry/data manipulation positions, because of the volume of claim data that comes in, much of which is still on paper. I recently worked with a small company that had four data analyst. One processed 9 data streams, comprising about 70% of their total claim volume, the other three managed to do one stream apiece. He had cornered himself by being literally too productive to promote to more interesting work.
Several months ago, I worked with a small marketing firm. They had a fairly large predictive analytics team, but all of their production models had come from a single modeler. The other modelers were building quite a lot, but none of them were winning out to production (their process, which is pretty common, was to run several models in parallel and move the best performing model to a production environment).
That actually sounds pretty fascinating. Want to share more details?
Here are a few examples off the top of my head- One of my first consulting jobs ever was to reverse engineer a bunch of models a former employee of a small engineering company had built. The company’s IT was atrocious, and to work around them, the employee in question had implemented a surprising amount of code in sql queries that were embedded in excel sheets (in general, the ways people work around bad IT make the situation even worse). These models and spreadsheets had become part of nearly every member of her team’s workflow. She left the company, productivity ground to a halt (with no one to keep the spreadsheets up to date, they were suddenly at the mercy of their terrible IT) and they spent a small fortune in consultants to get things back to where they were.
Insurance companies have a lot of data entry/data manipulation positions, because of the volume of claim data that comes in, much of which is still on paper. I recently worked with a small company that had four data analyst. One processed 9 data streams, comprising about 70% of their total claim volume, the other three managed to do one stream apiece. He had cornered himself by being literally too productive to promote to more interesting work.
Several months ago, I worked with a small marketing firm. They had a fairly large predictive analytics team, but all of their production models had come from a single modeler. The other modelers were building quite a lot, but none of them were winning out to production (their process, which is pretty common, was to run several models in parallel and move the best performing model to a production environment).