As a database professional I have danced back and forth between operations and development. Through my experiences I've worked with different aspects of the SQL Server technologies and even some open source platforms like MySql as a database and R for analytics.
One thing I have learned from these experiences is that data mining is sometimes just that - "mining". My introduction to analytics involved a personal side project analyzing some consumer data. For this project I downloaded MySQL and installed workbench, I worked with IT staff to get a VPN connection and fired up some analytical tools.
After getting situated I recall grinding out queries and classification algorithms for several hours. I could do some basic classification and cherry pick some whale customers. However, try as I might, when I looked at predictive algorithms, my only revelation was that all the orders from the state of "California" would be coming from the country of "United States of America".
Now I will fully admit that I was not and still am not a statistician. The whole project was a nice stretch to see if I wanted to try on a Data Scientist's hat. I was most certainly punching above my weight and may not have understood the tools I was using or what I was doing. In my defense there may not have been enough data there to make earth shattering insights. One customer database with limited data - demographic or otherwise may have hamstrung me.
Regardless of the reason, at that time, I was more or less a failure at sophisticated analytics. What this experiment has taught me is that I am better suited to outfitting those analysts with the data they need to go digging.
During gold rushes (Californian or otherwise) there usually a huge influx of prospectors set out to go mine some new found deposit. Some strike it rich while many, many others end up finding nothing. These situations are a risky proposition.
The safe bet is to be an outfitter. While every Tom, Dick and Harry are heading into the hills to make their fortune, there are a few outfitters that are selling the picks, mules, and provisions for those guys to do the exploratory work ahead.
Since my side project, I have come to understand and used SQL Server Analysis Services with success. However, I am still most comfortable building the integrations and back end data structures for the analysis. While I may not be the best analyst on the mountain I can get you a warehouse, some BIML generated SSIS packages, or a donkey if that is what you need to get the job done, and that’s OK with me.