“It was a split-second decision.”

How many times have you heard those words or used them yourself without giving them another thought? Today, let’s consider the significance of this familiar phrase. It suggests a mental summation so compelling and yet so quick that we’re unable to identify all the contributing factors. Like a “hunch,” the phrase describes a decision process that takes place beneath the level of our conscious awareness, as if from a black box.

In this accelerated and impatient age, when marketers have a brief window of opportunity to capture the interest of online customers, sometimes it seems like every situation requires a split-second decision. Presenting a product or service unsuited to the online visitor, or taking an exasperating few moments to present the right one, will lead to high bounce rates and lost conversion opportunities. Consumers are restless, and if the right information is not there the first time, they are quick to go elsewhere. They will rapidly surmise, from very few cues, how likely a site will be to satisfy their needs.

Recently, Brad Rencher, Adobe’s Senior Vice President and General Manager of Digital Marketing, called for marketers to adjust to how “the marketing world has become more complex, sophisticated, and fast-moving,” resulting in the need to address the challenge of “last millisecond” marketing. To create the kinds of digital experiences that engage consumers and drive them to act, brands must deliver a personalized experience in that instant between an action—a click, a tap, a swipe—and the next step in the consumer’s journey. The difficulty of achieving this is certainly familiar to those of us who think about how our brand presents to different psychographic segments, or about which offers would be most compelling to individuals based on their unique purchase history, or about any other number of experiences we might target to individuals based on the likelihood of their driving consumer behavior. As clear as the opportunity is if marketers can get it right, it’s equally evident that the challenges to achieving this success are significant.

In fact, it’s only in the last few years that marketers have gained access to technology powerful enough to take on this challenge and aspire to make the most of each moment a visitor spends onsite. Computers and networks are now fast enough to evaluate the available data and make the best decision on what to present to a consumer, in a timespan that is barely perceptible. Moreover, sophisticated machine learning algorithms can ensure that over time, marketers are able to anticipate, and successfully respond to, visitor’s interests with ever-increasing accuracy.

The recent release of Adobe Target Premium provides marketers with the latest data science technology and algorithms to deliver automated, personalized results at the last millisecond to a multitude of devices, over a variety of bandwidths and channels, wherever the customer may be. Specifically, Target’s automated personalization capability addresses the challenge of how to determine and deliver the right, targeted experience to each consumer, based on everything known about that individual at the point of interaction.

The figure below illustrates the basic steps of the automated personalization algorithm. First, Target gathers all relevant data about a consumer for processing and analysis. Each time consumers visit a social media page, display ad, or website, they generate potentially useful data. Automated personalization will sift through the data and learn over time which variables are most predictive of how a consumer will respond to different experiences.


Using on a wide variety of these predictive factors, it generates a “user-based score” that represents the likelihood of a response to an experience. Then, a “general score,” reflective of how the overall population is responding to different experiences, is combined with the individual’s user-based score to capture the right weighting between the individual’s past behavior, and the “wisdom of the crowd” in predicting how that person might respond to different experiences in the future. Finally, a machine learning approach known as the “multi-armed bandit” ensures that automated personalization continues to test and learn on an ongoing basis.


Those learnings are regularly pushed out across Target’s global server network, where the rules for the user-based score are updated every six hours and the rules for the general score are updated every 10 minutes. This ensures an ongoing cycle through the fundamentals of last millisecond marketing: 1) listen (gather data), 2) analyze and predict, 3) make decisions, and 4) deliver a personalized message, at scale and with a split-second speed that can only be achieved through the most cutting-edge technology.

Of course, in the end, all the data science and computing power in the world won’t make a difference if the product sits idly on your shelf. That’s why automated personalization was designed with usability and marketer-friendliness as defining features. Ultimately, the technology is just an enabler. The companies that get the most out of it will be those with savvy marketers who evaluate the results, refine the strategy, and adjust the implementation to better connect with consumers and drive business objectives. To that end, Target offers a workflow that makes it even easier to set up and run one of these activities than it is to create an A/B test, and includes reporting capabilities that highlight the insights generated by the underlying mathematical models.

To learn more about automated personalization in Adobe Target, or to contact us, visit our site.