Behavior vs Demographic clusters

Behavior vs Demographic segmentation is and was a hot topic since the first websites came along. Since I remember companies use to segment their markets related with Demographic criterions, which means dividing markets into groups based on agegenderincomeoccupationreligionrace, etc. This criterions divided into Hard, Medium and Soft. The Hard ones are those that normally never change like the birthdate or gender. Medium are those that can chance from time to time like single, married or divorced. Finally, the soft ones are those that can change any time.
As you can image most of the variables are soft. Actually, inside the soft variables you find the those related with “moments” in people’s life which are the most useful for conversion matters.
Based on the Meta Analytics model, systems (companies) as well as people are living systems in constant change. That is why most of the variables that we can use for clustering are in constant change which means that unless you can connect with people’s brain you won’t have the chance to get all that information for its analysis.

Behavior segments are those generated based on what people do.

People is constantly doing things and a big part of them (if in Internet) can be measure.
So, let’s think about a well known situation. Your company has a newsletter and you wanted to use your newsletter with two objectives, increase loyalty and get incomes. If you use a Web Analytics platform that allows you measure based on each user and it’s userid (like Yahoo Web Analytics just for mention a non paid solution). You can use the platform API, connect it to the email marketing solution, query the Web Analytics Database to get users that yesterday where navigating a particular product, or in case of a content site, a particular content and then send the newsletter. This process can be automatic which means that people will be receiving a newsletter with information that is relevant to them, because just a day or two before they where navigating those contents. What do you think? The results are incredible high lifts in all the performance and loyalty metrics.

The other interesting point about Behavioral segmentation is that allows you to identify the relation between variables (Anova*) in “real time” so you can achieve your objectives in a more effective way even if you have no idea of your user gender, age or country.
Finally I’m not saying “Just use Behavioral” what I’m saying is do not stock on your actions just for not having a demographic information. If you have it, just use it as another variable of analysis and try to identify if there is any relation between a demographic variable and the expected result or conversion.

* The Anova model (Analysis of Variance) is a collection of statistical models, and their associated procedures, in which the observed variance in a particular variable is partitioned into components attributable to different sources of variation (Wikipedia)

There are three classes of ANOVA models:

  1. Fixed-effects models assume that the data came from normal populations which may differ only in their means. (Model 1)
  2. Random effects models assume that the data describe a hierarchy of different populations whose differences are constrained by the hierarchy. (Model 2)
  3. Mixed-effect models describe the situations where both fixed and random effects are present. (Model 3)

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.