Vantage point




Thursday, May 29, 2008

A question for marketing analytics folks

In the last few years, marketing analytics has taken off in a big way in India. It is the relatively little known outsourcing success story. Firms from the US send their data to India for sophisticated analysis, specific to marketing questions, and people in India, typically MBAs, analyze the data and report findings. Marketics is a well known firm in the field, with HP Analytics and Dell Analytics also doing a fair bit of business that way.

A few of my friends work in those firms, and I have often been curious about the statistical packages they use for the analysis. My research also involves wrestling with marketing data, and I have used several packages over the last couple of years. In the rudimentary polling I have done, it seems like SPSS, SAS, Lisrel, MPlus, Minitab etc are the popular packages. A few of them, very few, use MATLAB. All of these are proprietary, and no doubt expensive, especially for companies.

Here is my question. I have yet to meet a person from the marketing analytics industry who uses R. They also did not seem to be aware of others in their firm using R. I wonder why that is. It is a very powerful package, which can do everything that the aforementioned packages can do, and more. It is a lot faster. Now ideally of course, you would want to use C or C++, but that involves programming which the analysts might not feel comfortable doing. R does not have that problem either. It is not as "black box" as other packages, but it is still fairly simple.

Plus, R is open source, hence free. The R community is also very active, so just one google search can help you find packages written by others which can allow you to do everything, from complex regressions to cluster analysis to structural equation modeling, to segmentation, to.... literally anything that a marketing analyst would want to do. And did I mention it is FREE? Yes, I did.

So why isn't R being used as extensively in the marketing analytics industry? Has it even been tried? Have the bosses or senior analysts even thought of switching to R? Is it because the salespeople for proprietary packages have been able to wield more influence than they should? Or is there some other reason for sticking to bulky and proprietary packages that I am missing?