Monthly Archives: June 2011

  • 0

Profiling a potential client

Recently someone asked me a reasonably straight forward question. How do we profile potential clients based on the input from a simple web form? I gave some suggestions to improve the form so that our intuitive profiling becomes a lot better. The person asked me if there is any way to automate this process at least partially. Well…May be!

Customer profiling has become extremely important in today’s customer-oriented business environment and is a well characterised specialty. Various statistical techniques like clustering, neural network algorithms and logistic regressions have been used for customer profiling. But in this case the problem is even more complicated as we are talking about profiling of potential customers.

Still a neural network predictive model based on past experience may still be the right answer even in this case. But how much data we can collect from a simple web form? Not much.

Is it better to include open ended questions so that we can collect as much data as possible? Open ended questions may not be amenable for model creation.

It may also be important to collect and analyse all the information available during the HTTP request, that the client did not enter personally. Like the IP (and hence the location and sometimes the employer), browser type, OS, referring URL etc.

Is there any way to analyse free form or open ended form data? More on this next time.