Data Analysis in direlights

    Hey, we’re back again with another weekly blog post! This time I want to talk more about market research and the ability to make business decisions based on qualitative, quantitative data instead of gutshot feelings.

    Picture a really large company; let's take Google as an example. Google has tens of thousands of employees, all living in different cities, different conditions, working on different projects with different end goals. And all of these people are interconnected; making a business decision on one project could have a silent yet massive impact on a completely different project. For this reason, one person can’t be making decisions randomly based on what they feel is right; every decision is reviewed by multiple people, and the root cause for the change has to have a foundation in solid data rather than a “I think this would be a good idea!”.
    This is why data science is such an important role in modern day business; all of the hundreds of decisions made daily require data to be certain that a given decision will result in a net cost benefit for the company. 
    And, I’m trying to bring a (albeit shrunk down) version of this to Direlights! We’re going to be sending out an interest form soon for merch, candle scents, and other assorted information that we believe will be helpful to Direlights for our business decisions. That’s the data + market research part, but there’s also a cooler second part. In consumer research, there’s an idea called AB testing; sometimes it also goes by ABZ testing. These are functionally the same thing, but ABZ testing has one more option.
    But option for what exactly? Well, I’m glad you asked, fictional reader! We were brainstorming ideas for our interest form, and we were considering having a small reward for the people who gave us the most helpful responses, in the hopes that it would drive more engagement with the form and give us higher quality data. But, we have no data to support this theory; it seems like a good idea that would work, but we cannot analyze the resulting data effectively. This is where AB(Z) testing comes in; in one form, we advertise the fact that the top few responders will get a free candle, and in the other, it’s just a normal interest form. Then, we send the two different types of the form out to half our testing population respectively, and boom! We have a control group (no candle advertising) and our experimental group (candle advertising). This way, we can see if the data difference we get with the experimental group is of enough of a higher quality that it justifies spending the money on the extra candles. 
    So respond to the interest form! Maybe you can get a free candle – or other goodies ;)

Written by Odin Nichols