Dr. Punj, is your analytics focus part of an IT or business curriculum?
It's part of our marketing curriculum. I've found there are IT curricula that focus on analytics but what I think is the greater need is for chief marketing officers or people doing marketing strategy to get deeper into analytics. Until now, the domain of business analytics has mainly been in IT under the CIO. I see this transformation occurring gradually, but I would like my students to be ready.
And your graduate students are taking graduate tracks on their own time and money?
Yes. These students are all working at corporations, mostly private sector in the greater metro NY area so we have students with jobs in services, health care, a whole lot of insurance and financial institutions and banking, even consumer product companies. Most are full-time middle managers seeking career enhancement or better job opportunities within the companies they are working at.
Are they arriving with IT or marketing backgrounds or both?
They come from all over. Some have come from IT seeking to make a transition into marketing.
We’ve seen evidence that chief marketing officers will be assuming more of the technology decisions formerly held by the CIO. Does that make sense to you?
I think it might be accurate to call the next generation of chief marketing officers the people with business analytic skills that previous generations of CMOs lacked. They are going to be software savvy and predictive modeling savvy. There will be less reliance on CIOs. We can see that it's not unhealthy for CMOs to become more self-reliant on their IT needs just because of their backgrounds and skills and the people working with them.
How do analytics get into the learning process compared to the way marketing has historically been taught?
The history of marketing research goes back almost 50 years, is very statistically oriented and has ties to political science research. Not much has changed in marketing research in the last 30 or 40 years and it is no longer satisfying the needs of companies. In market research the data is usually attitudinal -- as opposed to behavioral -- and based on surveys. You'd call people and ask what they think of a product or service, what they bought, etc. That's old hat because people no longer have time or inclination to respond to phone or email surveys.
And analytics are now applied to data in behavioral and transactional measures?
Yes, the reliance on attitudinal data we once had in market research is fading and being replaced by behavioral data, what people are buying online and their interactions with call centers. The new phenomenon people are calling 'big data' is 80 or 90 percent of the behavioral kind. This means that some of the attitudinal stuff on which marketing research is based is no longer sufficient for companies to make good decisions based on what their customers do.
Are students and academics ahead of this transition?
Students are realizing their skills need to be more up to date, but I think that transition is still lacking in academia. We know academia typically lags the business world by 10 years. What I am doing, with some support from IBM, is making the transition to marketing analytics. For instance, we have both marketing research and analytics courses this spring and students are voting for analytics with their feet. I have twice the enrollment there as I do in research. Students are reading about jobs, seeing what their competitors are doing.
How have you worked with IBM and what is the benefit for you?
The support from IBM has been access to their software, specifically SPSS Premium Modeler, the product that evolved out of SPSS. That has been advanced over the years and has about 22 algorithms of various types we use. My own relationship with SPSS began in 2003, but continued after IBM bought them. As the tool has grown, the focus is not for students to learn algorithms, it is to solve business problems. When a provider like SPSS builds an algorithm to do it, that's how it enters the picture.
How immersive is the math and statistics versus the technology you are using in your classes?
Students need to know the inputs that are needed, and how to interpret the outputs. So between the input and output a lot of black box stuff is occurring and a lot of complicated and detailed math. We don't get into that, and you would need a master's degree in computer science to decode what's going on there. We want them to have an idea of what's behind it. If we are discussing neural networks I'll give students an idea of what that is and how it works conceptually, but we’re not writing equations.
Is it enough to tell a future CMO they will need a level of statistical understanding to know what that black box is up to?
They should have an idea of conditional and joint probabilities, some math and statistics, but not much beyond that. This goes back to the discussion of CMOs becoming self-reliant. A CMO is not going to or be expected to write a complicated mathematical expression, but they should know enough that if the output or the result from a neural network analysis comes across their desk, they should know what it means.
Are these predictive analytics like those we're used to for measuring things like lifetime customer value or propensity to churn?
Churn and lifetime value are important concepts but I think the field has moved beyond just those things. There is a new focus on unstructured content we get into, also social media, Twitter, Facebook and all the other verbatim data available. My ultimate goal is some kind of analytics that extracts, on an incremental basis, all the information it can from unstructured data and combines that with the information from structured data. That way you get some kind of internal validation that through people’s behavior you have found some trend and validate that through what they are saying. You are adding predictive power to your research.
What other types of marketing problems do the analytics address?
We use predictive analytics and predictive modeling of every kind. I cover artificial intelligence techniques, genetic algorithms -- things not typically found in marketing. It's applied to solve real problems like market basket analysis. It's very difficult to use any statistical technique to do market basket analysis, but an algorithm makes that much easier than doing it statistically. The focus is not on technique, it's on solving a problem. For instance, if your company markets a laptop and a printer, what will your customer buy next?
Are tools like SPSS accessible to a wider and less statistical audience now
Yes, certainly a lot of IBM commercial customers are using this, and among academics there are four or five U.S. institutions using the same software package. From their Information on Demand conference I see IBM is also interfacing elements of SPSS in Cognos and Netezza software, so people who like to use that software can stay with it and use advanced analytics. I also see other tools like SAS; their statistical software gets used a lot in insurance and banking and areas that are very actuary driven, but we don't see it as much in marketing.
We see the corporate demand for analytic skills growing, do your peers in academics also see an urgent need?
We are aware, but like I said, students are voting with their feet. I often refer to a McKinsey Global Institute report last year that suggested we are going to face a shortage of 140,000 to 190,000 people with these analytic skills. Students are sensing this. It's self-motivation and self-preservation in unsure economic times. They are reading industry reports and sensing that they need these skills.