Your testing and targeting solution is hungry. Its appetite for meaningful data is limitless and, like a teenager, the more relevant data it digests, the more productive growth it can achieve more quickly, and with greater clarity. Without rich data to crunch down on, it gets tired and cranky, spitting out assumptions based upon whatever “average” analysis it’s able to achieve using limited out-of-the-box segments or rigid “black box” decisioning systems. These assumptions are created with algorithms running on basic first-party data utilizing limited, variance-ridden analytics data.

This low-data fuel, like low blood sugar, is dangerous for the optimization practitioner, who can quickly latch on to basic revenue lift resulting from high population or basic segment levels (e.g., new vs. return visitor). It’s easy to push out this broad winning content and call it a success, but what are you missing within this high-level population or basic segment view?

The danger of acting on basic data assumptions is that you get only an approximate view of a diverse testing population, or “false average customer,” who in actuality doesn’t exist. What you’re missing is a more granular view of the customer. For instance, is this person prequalified for this offer based on geography, demographics, or even third-party data sources like credit score? Is the customer a combination of profile parameters, like geography with referrer variables, that identifies the most predictive segment to target for greatest return? Perhaps basic segmentation is masking a more predictive variable or compound audience segment beneath the surface that won’t be analyzed or uncovered unless this data is fed to the solution.

Adobe Target provides custom segmentation for filtering your results by granular audience segments to accurately target them for even greater exponential return on investment (ROI). This is why our out-of-the-box segments are customizable with parameters relative to any data source you feed into the solution via a strong set of application programming interfaces (APIs). There’s even a batch profile upload API for customers who wish to bring in updated visitor profile data augmented with their offline branch, call center, and third-party data sources. These can be stitched together within a data warehouse, customer relationship management (CRM) system, or wherever profile data is stored. This can be based on several levels of data points, or even using a data management platform like Audience Manager to regularly update the data. Of course, the master marketing profile, which unifies all profile data collected within the Adobe Marketing Cloud, combines all of your digital first-party data into one profile for ease of aggregation within your data store.

A common question I hear from diverse companies across industries and Target customers is: What about Big Data? What if I have years of historical customer data that I would like to use for personalization of my content? Previously I mentioned feeding the “right data” to your testing and targeting solution. Per my analogy, you shouldn’t feed a growing, hungry teenager a bunch of junk food. We have extremely advanced customers in the retail, travel, financial services, and other industries with large, detailed profile data on their customers. This is important, especially considering the frequency of travel in the travel and hospitality space or new investments in the financial space. Feeding two years of data to a solution is valuable, but it can also bog it down. Only through testing and filtering by segments can you define the “healthy food,” or most relevant data, to feed the system.

This topic was discussed in a recent panel in which my colleague Gina Casagrande discussed methods for refining your data and targeting the most effective variables for personalization.  Manual testing and rules-based targeting, filtering reports by levels of variables, and comparing performance illuminates where targeting opportunities make sense and where they do not, defining the most valuable, or “healthy,” profile data to feed your solution. Analytics can assist with this detailed view. For customers of Adobe Analytics and Adobe Target, advanced data synchronization between the two solutions brings broader, detailed context to your analyses. This allows you to apply any Analytics audience segment and success metrics to your test results for unlimited drill-down and identification of opportunities, even for segments or metrics built in after the test is run. This speeds up the process of identifying the best opportunities for targeting relevant content based on your test hypothesis, and what “junk food” data is irrelevant and not healthy to feed the solution.

Oftentimes, a testing and targeting solution has a single practitioner or small team that may need assistance identifying the right data. They need help with segment discovery or uncovering the most predictive variables and combination of profile variables to target. Although automation can  be mistakenly seen as an advanced capability, it is invaluable in initially identifying where to focus your testing and personalization efforts.  Don’t be fooled by optimization point solutions that offer “segment discovery” by looking at disparate tests and surfacing segments that react significantly. All of these tests focus on different hypotheses and areas, which means aggregate or high-level analysis again results in broad and unqualified assumptions about significant segments. How are they reacting and why? Can we draw parallels between all of the hypotheses within all of your tests?

What is needed is automated personalization, which is unique to Target. Target leverages a set of self-optimizing machine-learning algorithms to automate the delivery of personalized experiences and content to the individual. It also provides a detailed insights report that evaluates visitor profile data and surfaces the most predictive variables—segment discovery—relative to the location and content fed into it. It also helps to discard the non-predictive profile variables based upon success metrics or conversion goals that mean success in the locations where it is run. Like having an automated nutritionist, the algorithms will identify the healthiest data to feed the system for targeting at different locations of your digital channels. They will also identify which variables you should exploit for segmentation and targeting to remain productive and healthy in terms of program delivery and growth. The immediate success of this approach was discussed in a recent blog by my colleague Kevin Lindsay on the payoff of effective personalization for an organization.

So, allow Adobe Target to help you grow your program with its open source APIs, automated personalization, and self-regulation to assist with identifying the healthiest data to feed your growing program, and see how you can produce the most profitable, accurate opportunities for targeting and personalization across your many digital channels and touch points.