31 January 2017

Day-to-Day Decision Making with BI – The Forgotten Scenarios

Business Intelligence is just a coined expression to distinguish a set of concepts, methods and technologies built many years ago from other Decision Support Systems. So let us forget about all of those too many fancy words for a moment. They are all mainly used with the same goal:

> Decision making supported by technologies.

The idea can be very simple: too much data for a "normal" human to comprehend and gather knowledge from, without the help of computers. So by now you are thinking about a big enterprise with thousands of workers with computer systems recording all their financial, human resources, supply chain, customer relation (etc.) records. This is where Business Intelligence technologies come handy, to make sense of it all.

Well, we are not talking about that today. Decision making happens every day in our professional lives that affect directly the company you are working with, even the smallest ones.

The coffee shillyshally

[Should I go for a coffee break now or later?] I know, you must be thinking I am crazy, but it is just an example as any other to explain you simple and then go for a little bit bigger. Imagine that indeed you start manually collecting every day the time you go for this break and at the end of the day you classify manually your days’ productivity into a categorical output [bad | so-so | good]. At the end of the year you will have a couple hundreds of records that with a little bit of luck will tell you that there is a correlation between deciding to go for a coffee break between 2pm-3pm and a good daily productivity. Next year I am starting to have those breaks whenever I can around 2pm…

You see that little brain named [Knowledge] above? That is the machine working for you, even if it would be easy for a person with some statistical skills, they could just use some easy to understand tools available online for free nowadays, for example MS Azure Machine Learning. In minutes you could load your Excel sheet and get an output from an analysis method you choose, without having to do the math yourself.

The trophy hunting

You might agree that a lot of companies’ success is very dependent on the people working for them. So it makes a lot of sense for them to hire the best human resources possible. Let us take this context as an example.

[Should I hire this candidate or not?] The recruiter will be backed up with years of experience/knowledge on the field when about to make this decision, but could he get some more support? Put yourself in the position of the recruiter for a moment, and by the way, you do not have much more than your old friend excel with you. You could translate those years of experience/knowledge to build your own framework with what makes sense to evaluate and categorize in a candidate. You will manually record that input for every candidate but you cannot make much sense out of it until you know this candidate, now transformed into an employee, is really a good asset for that company or not. Most companies should have at least annual Performance Evaluations, and if by some reason they do not, this probably means that besides recruiting, your job should also be to track the performance of these candidates/employees.

Remember, you still have your Excel with you so that is not the end of the world. After a couple of years, you finally have enough juice (data) from both sides [candidate | employee] so you can start your analysis and try to find correlations between your candidate profile and respective employee profile. Joining those two sides (by the person’s unique identification number) will get you a dataset ready for input on free online available tools like the previously mentioned MS Azure Machine Learning. When I talked about making your own framework to evaluate the candidate I meant that you can even come up with a category named [Gut Feeling] where you classify that candidate interview with your Gut, but that is just me talking and my incompetence on this field. But in the end I would like to see if my Gut was telling the truth or not by checking if there is a correlation between my belly and the years of performance to come from that employee. If there is no correlation between those (and I assume that it will not since I am not an experienced recruiter), maybe I should be looking at some other categories I have input on the candidate profile.

The process is the same for every scenario you find on your Day-to-day decision making, letting the machine do the little [Knowledge] brains’ work and use these insights for the next time you will be about to make that decision again.





      André Correia