Customer AI in Real-Time Customer Data Platform

Transcript

Luma, an athletic apparel retailer recently launched a new line - of product, LumaSmart Hoshi. Launching a new product can be costly, so it’s critical that they - market it efficiently. We’re going to show you how - Luma harnesses the power of AI by using Adobe’s Intelligent Services, to identify the right audience - to target with customer AI, and we’ll show you how they can do this all without needing a - team of data scientists. Here in Adobe Experience Platform, Luma has access to marketer - friendly dashboards that surface important metrics and utilize up to date customer profiles for predictive insights and - downstream personalization. Luma ingest data from multiple sources, including their CRM, call - center data, loyalty rewards, and point of sale data. All of it is stitched together here in Adobe Experience Platform with their appropriate - data governance in place. The data is then mapped to the consumer experience event Schema, which makes it easy for AI - and ML to scale with accuracy because it can understand, train, and deploy predictive - insights on a common data set with common language - reducing the amount of time for data preparation. This data across both - Adobe and non-Adobe sources is the foundation for powering - intelligence services. Here we see the two intelligence - services now available, Attribution AI and Customer AI. Let’s create a new - instance to identify people who are likely to purchase - LumaSmart Hoshi leggings. Using customer AI, a marketer can easily configure - their desired predictions without data science expertise. It’s a simple three step process. First, choose whether you want to predict propensity to convert or turn. We’ll choose the data - set for this prediction that we’ve already prepared a map. Based on Luma’s business goal, we could define a subset of the population to run this prediction for. Next, we define a goal - on the propensity score consumers will have to - buy women’s leggings. We can set the timeframe that - we want this prediction for. And last, we define - how we want the results or score to be generated. For example, we can select - weekly at midnight on Sundays. Luma can exclude any - events from this prediction that may be an anomaly and skew results. For example, if the website was down and we can omit that data point and voila, the setup is complete. Now, customer AI is hard at work to provide intelligence and - score each individual profile on the likelihood to purchase - LumaSmart Hoshi leggings. After training and - scoring using customer AI, the predictive scores are written back to experience platform on - each individual profile level. And we can also see the - predictive scores in aggregate through this reporting dashboard. Here we see the high, - medium and low distribution of propensity scores - across Luma’s consumer base and the top influential - factors or reasons why. They can provide richer context that we can use for more - granular segmentation. Luma can download the raw - scores through the APIs to get these insights at - the most granular level to power custom dashboards - through BI tools like Microsoft’s Power - BI Tableau, or Looker. We can also activate these insights across Adobe Experience Cloud - or non-Adobe applications like call centers. For brands like luma that - have Adobe Experience Platform or a realtime CDP, Adobe provides a seamless - integration with customer AI to further accelerate - insights into action. We click create segment here. I can activate this audience - of high propensity users with access to over a - hundred destinations. The logic is already applied and we can make any relevant - updates to the segment. We’ll give it a name and click save. Let’s take a look at a - real time customer profile where we can see all the - information we have on our users, including their propensity scores. A score is generated for each - and every individual prospect and customer for Luma. That’s an AI score that you’d often need a team of data scientists to provide. But with customer AI, it’s baked into the - Adobe Experience Platform and ready for the marketer - to take action on. Now that we have this high value audience, let’s take a look at - how we’re able to power a personalized customer - journey to Luma’s consumers with this intelligence - across social, email, and on the website. First, let’s start with social. Using Adobe’s Real-time CDP, we can activate the segment to Facebook for social targeting. Here, we select Facebook as a destination for our new audience. And now, we can target Luma users who have a high propensity - to purchase Hoshi’s leggings while they’re browsing Facebook. Here in Facebook, luma can see their Adobe Experience Platform - audiences they’re targeting. Our high propensity for - Hoshi leggings audience is already visible and can - we used in paid campaigns. We’re using this audience to create a paid retargeting campaign with a business objective - of driving visitors with a high propensity - to buy the Hoshi leggings to the Luma website. Just like that, we can easily extend and increase the reach of - these high value customers on walled gardens like Facebook. Next, let’s see how we can target our high propensity users via email. To do this, we’ll use Journey Orchestration - application service which allows marketers to orchestrate individual customer journeys - at scale across channels by intelligently anticipating - every individual’s needs in real time. Here in Journey Orchestration, we can trigger a journey - when someone qualifies for this high propensity audience. So here in the journey map, we can see the segment trigger is set to our high propensity audience and we’ll send them an - email using Adobe Campaign. So when Sarah Rose an - existing Luma customer checks her email, she receives the Luma Hoshi legging email. Yet another channel where we’re able to reach - our high propensity audience and bring them to the site to explore our new Hoshi products. Last, we can personalize to - new and existing customers who have a high propensity score when they visit the - website using Adobe Target. The platform audience of high propensity to buy Hoshi leggings is right here in Adobe - Target’s audience library. The audiences in this library are a combination of those - we’ve created in Adobe Target and those that are shared - across the experience cloud from Adobe Experience Platform and other solutions like Adobe analytics or Audience Manager. Let’s take a look at the AB activity we set up using the visual - experience composer. Here you can see three - different experiences for the new Hoshi legging - hero banner on the Luma site. We’ve set up the test to target our high propensity audience, and they’ll get one of - these three experiences. In summary, we can see - how intelligence services could help us accurately identify - the right people to target with customer AI. With intelligent services, brands like Luma can scale and start turning - insights to action faster letting machine learning - take on the complexities, so marketers can work smarter not harder. Welcome to the next era of - experience intelligence. -