targeting imageAny system that uses rules to determine the content being displayed to a segment, persona or audience is a targeting system and not a personalization system. The goal of personalization is to surface the most relevant items to each customer based on their current needs and what you know about them as individuals. The more you know, the more you should be able to personalize the experience to drive higher engagement, conversion and lifetime value. Targeting isn’t bad; it’s just different from personalization. The following are some of the key differences between targeting and personalization.

Marketer vs. Machine Driven

Targeting rules tend to drive what the merchandiser or marketer thinks is important. At best, these rules are based on recent analytical data of big trends. While this can work in favor of a business pushing products with excess inventory, it can also have a negative effect if the targeting rules are based on marketer bias or incorrect buyer assumptions. Personalization engines, however use machine learning to focus on what is best to satisfy the customer’s needs based on behavioral data and known preferences.

Update Frequency and Complexity

Because targeting systems depend on people, they are slow to update. To determine that a targeting rule needs to be changed or added, a pattern must emerge in analytics or through an A/B test that is big enough for people to notice. The cause and effect has to be well understood. From there, new content may need to be created. Finally, rules are updated to show the new content to the new people or to update the existing rules. By the time the rules are finally changed, the trends that they were meant to capitalize on may also have changed. Conversely, personalization engines use machine learning to find patterns of behavior and continuously update the perception of each customer based on their current behavior, new data, etc. With a personalization engine, the content displayed to each customer is based on the most recent interactions from that customer and other customers. The engine adapts automatically to changing trends and new inputs.

 Macro vs. Micro Segments

Marketing segments or personas are usually limited to a handful of intuitive groups that the marketer can easily keep track of, like “sale shoppers” or “luxury buyers.” These groups are usually pretty large so they can be tracked and included in campaigns. Personalization engines take a different approach which can be thought of as dynamic micro-segmentation. In this approach, customers are continuously evaluated based on their known preferences and what they are doing now. Then they are compared to other customers with similar preferences, interests and attributes or purchases. The benefit of this approach is that customers are constantly re-evaluated by the machine and they may fit into more than one micro-segment at a time. For example, as their behavior shifts from self to gift shopping, the personalization shifts with them to show the most relevant content that will get them to engage and convert.

 The Human Cost

While targeting rules might seem easy to setup initially, they tend to become unwieldy over time. First, rules can quickly go out of date. What was valid today might not be in a week or two. Second, rules and rule conflicts are very hard to troubleshoot, especially for people who didn’t create them. So, if you’re a manager or the “new guy” trying to figure out the impact of changing rules that someone else created…watch out. The key to one-to-one personalization is automated learning. Marketing teams can “set it and forget it”; and the algorithms will keep up with changes in their catalog and content or changing tastes or trends of buyers. This is a much more agile and scalable way to match people with relevant content. More importantly, machine learning approaches are more granular, faster to adapt and don’t develop bias towards their prior conditions.


The goal of targeting and personalization is the same – to get away from a one-size fit all approach with stagnate conversion rates, to an approach that will show more relevant content and increase engagement, conversion and loyalty. By any KPI, automated personalization strategies will outperform targeting strategies. Targeting, which is a very limited approach that uses limited data, is also slow to adapt and can be loaded with bias. Automated personalization is a machine driven approach that is much more granular, much quicker to adapt and can remove much of the bias. It makes sense that automated personalization would outperform targeting. I have seen it over and over again.

Man & Machine – The Art and Science of Personalization

Don’t get me wrong, the machines are not taking over the world. Real world applications of personalization ultimately require both people and machines to get the digital experience right. Brand and domain experts understand a business and business constraints in a way that is difficult for machines to master. Brand executives and product managers and designers have expectations for the website look and feel. Ultimately what is displayed on the personalized website or application is the combination of human logic, human passion and machine learning automation. However, when you allow machines to solve problems that they are good at solving, like personalization, it can free up the creative people to do the work they love instead of managing rules.