ientation – it is not possible to judge the sort of risk return tradeoff that a user would prefer by studying their profile or by asking them directly. This information would be needed while recommending variable return policies (unit linked schemes), where the policy is more of an investment instrument that a life insurance policy. According to all the insurance agents who were consulted by this author, each socio-economic segment contains people with varying risk preferences. The agent’s approach is to offer a range of policies that the client can afford, and let the client choose on his own. However, over a period of time, the agent becomes aware of a user’s risk orientation by observing his buying behavior. Similarly, the web site should be able to ‘learn’ the user’s behavior by observing the user’s online behavior. A useful tool in this regard is web traffic analysis.The author suggests assigning a neutral risk category to each new user. At the same time, each cash value policy should also be rated in terms of the level of risk involved. The user’s rating should change according to the risk rating of the policies that are bought. Besides this, if the click through analysis indicates that the user is interested in policies with a certain type of risk profile, adequate changes should be made to the user’s profile. Apart from this, the click through analysis also helps in fine-tuning the profile of the user. It can also be used to detect changes in the profile of the user and to detect any life event that is occurring for the user.The three main objectives of the click through analysis are:Gathering and interpreting behavioral information.Fine tuning the existing profiles. Apart from being an aid in personalization of the site, web traffic analysis offers a mechanism for measuring the effectiveness of the web site.(see Appendix1) 9A LOGICAL ARCHITECTURE FOR PERSONALIZATIONA l...
