How Adobe Target Automated Personalization Shines
Adapting to Changes in Behavior
Visitor behavior on a site can change over time for a number of different reasons: for example the site may begin attracting a slightly different type of visitor, the competitive landscape may change, or seasonal effects may come into play. Adobe Target Automated Personalization adapts to these changes in visitor behavior, offering a huge advantage over the rules-based targeting of winners that results from A/B testing.
Because changes in visitor behavior over time can invalidate learnings from A/B testing and because these changes often go unnoticed for a substantial amount of time, those outdated A/B testing learnings continue to be applied as if they remain valid.
Automated personalization dynamically directs traffic to the experiences that have the highest expected conversion rates at any point in time rather than limiting its judgments to data collected over a limited testing period that occurred in the past, as A/B testing does.
Exploring and Exploiting Which Experiences to Offer
Adobe Target Automated Personalization features what’s known as a multi-armed bandit algorithm. The colorful name, incidentally, comes from the conundrum that a gambler faces when trying to decide which slot machine—the “one-armed bandit”—in which to toss money, and when it’s advantageous to move to another machine (and which one).
Initially, the gambler has no knowledge about the machines, so must make a tradeoff between “exploiting” a machine believed to have the highest payoff versus “exploring” the expected payoff that might come from other machines. Switching checkout lines in the grocery store in the hopes of getting checked out more quickly perhaps represents a more universally experienced example.
Similarly, the multi-armed bandit algorithm that Adobe Target Automated Personalization uses efficiently balances exploiting the experiences with known high conversion rates with exploring the conversion potential of other experiences. In addition, this type of algorithm ensures availability of current conversion data for all campaign experiences and takes that data into account when determining the experience to offer. You can imagine that such an approach offers a huge advantage when limited conversion rate data is available—for example, at campaign launch or when adding an experience to an existing campaign.
Applying the Strengths of Two Models
In my previous post, I discussed the two different models that Adobe Target Automated Personalization applies for each experience (generalized and personalized). This approach lets the system leverage each model’s strengths to deliver superior conversion performance.
The system leverages the generalized models, which can be built with less data compared to the personalized models. This allows the generalized models to start serving new experiences faster than the personalized models. However, as the campaign runs and more data become available, targeting will become increasingly precise as the personalized models begin to have greater weight in determining the experiences to serve.
Empowering the Marketer
With Adobe Target Automated Personalization, marketers can conduct a detailed investigation to determine which visitor profile data points are the most predictive for conversion on a given experience. Such exploration can reveal valuable insights about the nature of the visitor interest in the current experiences as well as drive ideas for creative content to offer in new experiences.
For example, a marketer at a financial institution reviews the results of a campaign run through Adobe Target Automated Personalization and notices that Spanish-speaking visitors show high conversions on a specific experience. This observation leads the marketer to create new experiences in Spanish targeted specifically to this segment.
With Adobe Target Automated Personalization, marketers can uncover insights and respond to them by creating and offering increasingly targeted, relevant experiences.
Delivering Experiences with No Lag or Delay
When a visitor visits a web page, Adobe Target Automated personalization processes all available information about the visitor, identifies the experience most likely to lead to a conversion, and serves the visitor this experience. Research shows that you have up to one second for all this to happen or visitors start feeling frustrated and dissatisfied with the experience.
Adobe’s automated personalization solution handles all these steps in less than 300ms. Essentially, the visitor experiences no lag or delay from click to experience served. The extensive hardware infrastructure Adobe has in place in data centers all around the world make this processing speed possible by minimizing the distance the data has to travel no matter where in the world the visitor clicks.
Integration with Adobe Marketing Cloud
One of the biggest advantages of all Adobe digital marketing solutions, including Adobe Target, is its integration with Adobe Marketing Cloud. This integration enables three key advantages for Adobe Target Automated Personalization:
- Historical visitor behavioral data from Adobe Analytics can be easily used as predictive variables in Adobe Target Automated Personalization.
- Audiences (visitor segments) that marketers manage within Adobe Audience Manager are directly available for use in Adobe Target.
- Content managed through Adobe Experience Manager can be easily deployed in the targeted experiences that automated personalization serves.
In the beginning of this series, I stated that by serving the right experience to the right visitor at the right time, you can unlock the full potential of your visitor stream. Through this series of posts, I’ve tried to explain how Adobe Target Automated Personalization enables you to deliver these optimized customer experiences. First, by showing the advantages of automated personalization over manual A/B testing, next by explaining how visitor data fuels its personalization capabilities, and specifically how the Adobe Target Automated Personalization actually works. This latest post focuses on some of the key differentiators that, in my humble opinion, make the Adobe solution for automated personalization best in class.
Now that you see the potential that automated personalization offers, in a follow up post, I’ll give you specific, easy-to-follow steps for launching a project that lets you test the waters of automated personalization.