Behavioral Targeting is used to target visitors based upon their online behavior, such as category and product navigation, search keywords, and data entries in forms. This information is often ignored on websites today, yet can be highly predictive to drive offers and other content to website visitors. Adding other customer information about the visitor, such as products owned, propensity, or customer segment, can enable an even more personalized experience.
With Behavioral Targeting, an online visitor profile is created that includes the visitor’s current and past behavior. This profile may be used for segment targeting using rules or automated predictive campaigns.
A visitor profile would include online behavior such as:
- Visitor Activity – e.g. recency or frequency to the website, current or past conversion actions
- Entry Criteria – e.g. ad campaign, search provider, search keywords, date and time
- Session Behavior – e.g. last navigation clicks, page views, form input, site search keywords
- Environmental Criteria – e.g. IP address, geo location, mobile device, browser type, OS
- Customer History – e.g. customer propensity or next best product, customer value, products held, customer segment
Segmentation rules can be used as a way to direct A/B and multivariate tests, such as only applying a test to visitors using IE or Firefox browsers. Segmentation rules can also be used for targeting, serving up experiences based upon the visitor attributes.
Segmentation discovery can be done through the analysis of reports filtered by visitor attributes, and more sophisticated systems can automatically cluster segments to suggest segments that perform better with different experiences (combinations of variants).
Predictive Behavioral Targeting
The next stage of sophistication is using predictive campaigns to predict the right content to present to each individual visitor (as opposed to a persona or segment) that will yield the highest conversion rate. Predictive campaigns rely on a mathematical model that learns over time and dynamically adjusts to optimize results. Predictive campaigns are highly effective and easily maintained, since targeting is based upon a learning model rather than targeting rules. Rules may be used to constrain the model when necessary
Using predictive campaigns enables a mathematical model to automatically adjust the content displayed for a visitor based upon the visitor profile and content that yields the highest conversion rates. Though a self-learning model does not require rules, you can use rules to handle specific cases where you want to control the response and rely on the model for all other visitors.
Models are often used for determining the best offers to display on home pages or product category pages, enabling a broad range of offers to be dynamically presented to visitors based upon their most predictive attributes.