Visitor Data—the Fuel for Automated Personalization
The cornerstone of Adobe Target Automated Personalization is visitor data. That is what allows us to differentiate between visitors and predict their propensities for conversion. Of course, the performance of any machine learning system is only as good as the quality of the available data.
In our previous post about the power of visitor segments for automated personalization, we discussed an outdoor equipment retailer trying to determine which experience to serve its visitors—a Cycling experience or a Skiing Experience. In the example, we segmented visitors based on location and realized a large lift in conversion because we were able to show more relevant content.
However, visitor location is just one example of a predictive variable. Every time a visitor comes to your site there are hundreds of variables available to you, each with its own potential for segmentation. The question is how you most effectively take advantage of all this information. It was easy to analyze the segmented data based on one variable, but this task becomes much more complicated when adding more than just a few new variables.
Adobe Target Automated Personalization automatically identifies meaningful segments based on hundreds of out-of-the-box variables, including:
- Site behavior (e.g., new/returning visitor, search behavior, browsing behavior)
- Temporal (e.g., day of week, time of day)
- Referral (e.g., direct/bookmark, link from e-mail, search/display ad)
- Environmental (e.g., operating system, device, mobile/pc, browser).
The graphic below shows an example of variables in these categories for a particular visitor. I’ve also added a CRM category because with Adobe Target Automated Personalization, you can integrate additional data sources to provide for example data on past behavior. This data could include the dollar value of past purchases and the recency and frequency of theses purchases, both of which are typically highly predictive of future behavior.
But Adobe Target goes beyond these variables to improve the predictive power of automated personalization. Using advanced web-crawler technology to automatically segment the entire client website into Interest Areas, the algorithms take into account the areas of a website that a visitor has shown interest in. This can tell you a lot about a visitors preferences and interests. For example, with our outdoor equipment retailer, this technology might create interest areas for Cycling, Skiing, Camping, and Boating.
This automatic segmentation of the client site offers a tremendous advantage for large sites that host several business areas on one site. For example, visitors to Microsoft.com who spend the most time on the entertainment pages are likely to convert differently than visitors who are mostly interested in devices. This capability of Adobe Target Automated Personalization allows us to take this highly predictive information into account when deciding which experience to show to a given visitor.
The out-of-the-box variables with Adobe Target Automated Personalization can also be augmented with third-party segmentation data, offline sales information, and data from other customer-related sources, which allows automated personalization to get an even more refined and customized picture of the visitor.
One of the most powerful ways you can augment the out-of-the-box variables is by integrating Adobe Target Automated Personalization with Adobe Analytics (for web behavioral data) and your CRM system (for purchase history data).
After reading this post, you’ve seen how visitor data is the cornerstone of automated personalization. I’d like to challenge you to consider what data sources you could use with Adobe Target Automated Personalization to help it refine and customize the view of your customer so that it can serve the best possible experience.
In my next post, I’ll describe how Adobe Target Automated Personalization actually works, including the two models that it uses to deliver highly personalized experiences.