Today, at this very moment, there are e-business marketers and decision makers attempting to decide on a “website personalization” strategy, culminating in confusion and more questions than when they started. Unfortunately, the noise in the industry only adds to this chaotic environment, with every vendor claiming seemingly similar functionality.
Executives – who are eager to reach their customers and prospects in the manner most likely to be meaningful to those audiences – are torn between website personalization solutions that either enable them to manually create and manipulate rules or rely upon machines to automatically make choices on their behalf. “If we don’t start testing manually, how will we know what rules to create?” “If we opt for rules-based personalization, what will we do when we grow beyond what we can accomplish manually?” These are questions that plague many companies, and the intense arguments taking place between vendors are not helping e-businesses reach any answers. Truthfully, these are the wrong questions to ask.
Here is the right question: “How do we implement a website personalization strategy that fulfills our needs today and tomorrow?”
A Continuum That Begins With Rules-Based Personalization
At Amadesa, we believe the answer to that question hinges on a continuum offering. The web analytic offerings prevalent on virtually all websites today highlight basic foundational information about customers. Marketers leverage this data to determine what customers respond to so it can shape preliminary promotions. Rules-based personalization is great at using single data points to make decisions. A company in this situation might set several rules based on parameters such as IP address, gender and/or the search terms employed to reach the landing page.
For example, a particular organization might identify first-time visitors and offer them the chance to register for a newsletter. It might set a rule to reward frequent buyers, to target certain promotions to New York-based users, or to offer boots to someone hailing from a rainy climate. With this handful of rules, an individual or team at the company can decide what specific content should reach each identified segment (see screenshot with an example from Amadesa platform below). However, is one level of intuition replaced with another when rules are built to replace the strategy of testing?
At any rate, this is a personalization technique that works well with five or 10 or even 100 rules. However, once a business finds success with identifying behavior traits of various groups, it might wish to pursue more detailed segmentation. Perhaps a shoe retailer would like to test visual elements that appeal to West Coast residents, parents of girls, Canadians, men and women of every age group, etc. And once that e-business has more than 100 of these rule types live on a site, manual rules-based targeting is no longer realistic. It’s just not possible for human beings to track all of the possible combinations to accurately serve a large number of microsegments, nor can reporting provide a clear indication of success.
Moving Toward Microsegmentation with Automatic Personalization
This is where automatic personalization comes in. A solution based on a machine-learning algorithm can create tens of thousands of microsegments. These small groupings of visitor types are similar to those a business might identify manually, but they are created automatically and updated on the fly, every day, every hour, learning every time new data is presented. The hypothetical shoe retailer, for example, might identify microsegments based on season, geography, gender, income level, age, parental status, click-path analysis, past purchases, etc. Such microsegments supersede the value of rules-based segments, as they are virtually infinite and constantly evolving.
Of course, automatic personalization does not mean the machine runs the show. Marketers control the content served. Marketing and IT accumulate this knowledge to shape what is delivered to each microsegment. The smartest machines predict and confirm what the user wants, but also allow for options to override the automation to make a rules-based, manual decision when necessary. The e-business that has moved along the continuum from rules-based to automatic personalization has one outstanding need to fully realize the potential return on investment in its solution: feedback (also known as reporting and analytics).
Machine-based personalization has unlimited capabilities for creating targeted promotions, and the owner needs a way to view the results in a digestible, accessible manner. Clear reporting is essential and should deliver insight into what was served to each microsegment. For the shoe retailer in our above example, capable reporting mechanisms would enable the company to identify the top 10 or 20 microsegments and filter them based on specific data, such as segments of new visitors or customers who completed purchases in the past month. These reports affect every decision the retailer is likely to make, from procurement to marketing. Below is a sample report from Amadesa platform.
E-businesses questioning whether automatic personalization is better than rules-based or vice versa are asking themselves the wrong question. Instead, these organizations should examine how their needs are likely to change and whether their approach to personalization will enable them to move along the continuum from preliminary targeting efforts to tapping the potential of every possible segment. It is this continuum – along with a comprehensive feedback mechanism – that e-businesses at any stage should pursue.
