« Blog Home

Personalizing According to Persona

Posted on 6/14/2009 at 1:11am

One of our premier clients has recently proposed a personalization initiative for their flagship website. Our initial personalization discussions revolved around changing the site according to the user’s demographic data; once we know where someone lives or when their birthday occurs, or once they’ve answered a few simple questions for us, relaying information tailored specifically to their interests becomes a fairly straight-forward exercise. There’s just one problem: we have absolutely no way to access the user’s account information since it is stored and managed by a third-party vendor.

How can we personalize the site when we know absolutely nothing about the identity of the person who is using it?

As it turns out, by tracking the user’s behavior alone, we can capture enough valuable data to target the site to the user’s needs. Our strategy starts small:

  1. Using the site’s web analytics history, we identified a handful of behaviors that we are going to track (between 5-10 discrete actions). These mostly include navigating to particular pages or sections of the site.
  2. These different behaviors are aggregated into four or five clusters that represent user segments. These segments can be considered archetypical users or personas. For example, a “Value-conscious” segment is essentially a persona that performs four separate actions on the site.
  3. After we’ve identified our segments, we anonymously track a site visitor’s activity and quietly measure which segments they fall into.
  4. We change specific parts of the site according to a visitor’s current segment orientation, including marketing copy and callouts.

Anonymously segmenting users and changing the site accordingly is only one part of the problem; we also need to track the effectiveness of our behavior-driven site changes. Our approach is simple: leveraging the site’s existing analytics service, we’ll track which version of a callout was clicked (for multivariate testing) and the segment that was assigned to the user when they clicked it.

Our solution provides a personalization strategy that is relatively easy to implement and offers an incredible degree of flexibility to suit the needs of our client. Eventually, we expect that we’ll be blending our anonymous segmentation solution with a more robust package provided by vendors who specialize in data mining and recommendation engines. Ultimately, I foresee a personalization solution that provides expansive directed marketing efforts via the marriage of behavior and demographic data.


Leave a Comment