Behavioral Targeting -- follow up from Avinash

Avinash Kaushik has started an interesting discussion on Behavioral Targeting. Most of the time investing in Behavioral Targeting software and going forward with aggressive Behavioral Marketing needs a lot of skill set and indepth understanding of its use.

Premature use of BT Software may lead to various problems as a consequence of poor or no data, and no proper plan or skill to use the results.

The action based on the BT results can be the change in marketing source change (mode of advertising) or actualy change in website (content change, content location change etc). Each is a outcome of focussing something and discarding something. These strategic move may come by the dominance of customer segment by volume or quality.

You need to be aware for not doing one of these mistakes.

1. Loss of important but unprofitable segment: The segment that is found to be unprofitable (in direct calculation) may be profitable indirectly. They may be strong in spreading the news or good views about the company or product etc. They may even be a segment whom the other mass follow. Like for my own blog, major source of traffic is due to the inclusion of web analytics discussions while I am not a web analyst (but I learned it later from blogger friends and their books). There is huge chance that if I remove web analytics topics, I will lose much traffic. And this is not because that majority of the readers come for web analytics topics in my blog but the source from where they come are web analyst friends blogs referal links. (I could find this in the Google Analytics report.) In this scenerio, BT study may suggest something where the business benefit lies somewhere else, and the problem is not with the BT software but with the user (weaknesses listed above).

2. Over-reacting to profitable segment: This is as worse as above. If I like pink rose, it not necessarily mean I like a pink car. To find out accurately what customers want and what not, proper data is not enough but sufficient data is. This justifies Avinash's point 1.

3. Portfolio Moves ok, but how about market move: I am not sure how sophisticated are today's BT tools, but I expect them to fairly show the portfolio movement and give good segmentation picture. Testing of software and the method of using the BT software outcome can be done as Avinash has pointed out in his point 2. But the question lies, is that enough?

My take is, it is not. Both the methodology and the software need regular testing and upgrades. Market is moving faster than ever before, the technology changes, the macro-economic fundamentals change, and new products and process evolve almost everyday. This asks for us to be more prone to change than ever. We need a more advanced system in the likes of Enterprise Decision Management (the link follows James Taylor's blog where he writes on EDM), but I am not sure whether EDM is available for website management. If its not available, I am sure it will come sooner than later. In the absence of such system, we will need very skilled people and frequent testing infrastructure. Atleast once a month testing of software and the methodology is required (which I found practical and relevant while working for sub-prime loan market earlier). This is more important if the software vendor is not very proactive. I beg to differ with Avinash here.

This is all I have to say. And now its your turn to say. Please share your views in comments.