Use Customer AI Scores and Insights

Learn how to use Customer AI propensity scores and insights to convert and retain customers. For more information, please visit the Customer AI documentation.

Hi, I’m Hithala Trinidade, senior product manager. Let’s look at what kind of insights Customer AI generates, and how those insights can be used. Luma, an athletic apparel retailer, wants to increase sales of watches, and they’ve already created a customer AI instance to generate propensity scores. We will now deep dive into customer AI reports to see how Luma can use the insights and scores in the marketing. Intervention services are available from the Services link in the left navigation. Clicking on the Customer AI service card will show you all the instances created in the account. When a service instance is new, the hyperlink is disabled, as the insights have not been generated. You can click elsewhere on the row to open the right trail, and see additional details of the service such as name, description, scoring frequency, the prediction goal, and eligible population. After training and scoring have successfully done, the predictive scores are written back to Experience platform on each profile. We could look up any individual yield and customer profile, to see their propensity to purchase Luma watch, and we can also see the predictive scores in aggregate. Once you click on the service instance, the same details are available at the top, and are easily accessible via Show Mode. In the middle panel on the left, the scoring summary shows the total number of profiles scored, and categorized in low, medium, and high propensity buckets. The propensity buckets are data mined based on score range alone. That is, low is less than 24, medium is 25 to 74, and high is 75 and above. The score distribution on the right shows a visual summary of the population based on the score. The colors that you see here are based on the type of propensity score chosen in the configuration. Conversion are shown. If it’s a conversion propensity score, high scores are good, and we show them in green. However, if it’s a churn propensity score, high scores are bad, and we will show them in red. In the lower panel, we see the high, medium, and low distribution of scores, across Luma’s customer base. For each score bucket, we show the top 10 influential factors for that bucket. The influential factors give you additional details on why people belong to various score buckets. These score buckets and influential factors could be used for data analysis, customer segmentation, proselytization, experimentation, promotions, customer relations access management, product improvement. With these scores you can power customer segmentation, and targeting. Let’s create a segment from this high propensity users. The logic is already applied with the segment definition, and you can make any additional updates if you want. You can add additional conditions to the segment, like profile attributes, events, et cetera. We will give the segment a name, and description, and save it.
Now that we have this high value audience, we can activate it with both Adobe Application and non-Adobe applications. Since the propensity scores are returned to the individual profile, they are available in the Segment Builder, like any other profile attributes. When you navigate to the segment builder to create new segments. You will see all the various propensity scores under the customer’s name space, under customer AI.
At the individual profile level, Customer AI adds the percentile. This value provides information regarding the performance of a profile, related to other similarly-scored profiles. For example, a profile with a percentile rank of 99 for churn indicates that it is at a higher risk of churning compared to 99% of other profiles that we scored. Use percentile for segment creation when you want to target top X person of your population for marketing campaigns. Probability. This attribute is the true probability that a customer will achieve the predicted goal, within the defined time frame. When comparing outputs across different goals, it recommended that you consider probability over percentile or score. Probability should always be used when determining the average probability across the eligible population, as the probability will tend to be on the lower side for events that do not occur frequently. The score. Score is the relative likelihood a customer will achieve the predicted goal with the defined time frame. The value is not to be treated as a probability percentage, but rather, the likelihood of an individual compared to the overall population. Influential factors. These are predicted reasons why a profile is likely to convert, churn, or take the specified action. Factors are comprised of the following attributes. The code, which is the profile or the behavioral attribute which positively influences profile’s predicted score. The value of the profile, or behavioral attribute. The importance, which indicates the weight of the profile, or behavioral attribute, has on the predicted score. It ranges from low to medium to high. The output of customer AI can be used for personalization, targeting, et cetera, using Adobe Experience Cloud Applications, and services, and third party tools. Customers can create segments, leveraging the propensity scores within the Segment Builder, and these audiences will be available for using Adobe Advertising Cloud, Adobe Audience Manager, Adobe Campaign, and Adobe Target. Customer AI scores can also be uploaded in Adobe Analytics for exploratory data analysis. All real-time customer data platform customers will be able to create segments, leveraging the propensity scores, and activate them via destinations. Of course, the propensity scores can be uploaded into third party tools as well. So, you should now know how you can use customer AI propensity scores, and Insights to convert and retain customers. -