- Up next, I’m joined by Thiale Diop to go through the simple configuration and setup of the latest AI-driven and integrated tech across Customer Insights, that uses predictive AI to help you identify the customers that you may be at risk of losing, and automation of Dynamics 365 marketing to build targeted offers and personalized experiences to keep your customers coming back. So Thiale, thanks for joining us today.
- Thanks, it’s great to be on the show.
- So we’ve looked at the role Customer Insights plays in giving you a unified view of your customers when Satish Thomas was on the show. And for those of you new to Customer Insights, it helps you create a unified 360-degree view of your customers. You can bring in any data type using a variety of prebuilt or custom connectors, and this can be transaction history, point of sales data, behavioral data, customer preferences, product usage, or survey data, and much more. In fact, you can watch the show we did on Customer Insights at aka.ms/CIMechanics. And today, given the challenges of the past year, we really wanted to look at how you can combine, Dynamics 365 Marketing and Customer Insights to really protect and even grow your customer base.
- Right, and this is pretty topical because the past year has impacted a lot of traditional buying trends. As some people are now more comfortable buying online, to the point they might actually prefer it. And there are others who are transitioning back to buying in person. So, say you’re a business data analyst. Wouldn’t it be great if you can easily get a pulse on the relationship your customers have with the business at an aggregate or even per-customer level? And also from a marketing perspective, you never want to treat your customers like a faceless consumer. And so the more you can understand their preferences, the better you can serve them with personalized experiences, what we often call journeys, and tailor the communication so that each offer better resonates with your customers for stronger engagement. Like the one that you’re seeing here with an AI-generated personalized email, where we base the offer on their recent activity, or even offers that are triggered in real time by an event. For example, if they come into the store and use the store wifi, how can we make the most out of that experience?
- And that sounds really powerful, but what are the key pieces that you need to do to light all this up?
- Well, the good news is that everything is turnkey and ready to use. In our case, we’ll use Customer Insights with Dynamics 365 marketing, and I’ll walk you through just how easy this is to set up if you’re an analyst or a marketer. And I’ll also use an example from Contoso Coffee, which has both retail locations like cafes and an online coffee subscription service. I’ll start from the data analyst perspective. I’m in Customer Insights, and one of the things we give you out of the box are these prebuilt AI templates, such as the one you’re seeing here for customer churn, which is a predictive measure of our likelihood of losing a customer if we do nothing at all. This is all AI generated and then the risk is assessed based on several influential factors, such as the amount of time since their first purchase, how often they buy and much more. Now let’s configure this further. One of the things we really care about is a metric called lifetime value. If a customer has high lifetime value, we want to keep them coming back. So I’m going to create a combined segment of our highest customer lifetime value customers, called CLV, that are at the most risk of churn. I’ll be combining our churn prediction with our customer lifetime value that you can see here, also in Customer Insights. I’ve already created the churn prediction model, so now I’ll walk you through how to create the CLV prediction. I’m still here in intelligence and under predictions. In this case, I’ll choose the Customer Lifetime Value tile and click use model. To get started, I’ll give it a name. Let’s go with Customer Lifetime Value. And in the interest of time, I’ll walk you through the creation of a model I built earlier. The next step is to define the model preferences. I can choose just how far into the future I want to predict. And by default, the units are set to months. But as every business has its own set of nuances, for Contoso Coffee I’ve set this to five years. And I’ve kept the default for the model to calculate the purchase interval. And the model calculation will define the percentile of high value customers, but you can manually configure this as well. And we’ll see that in just a sec once the data analysis is complete. Next, you’ll see that I’ve added in my required data. In this case, POS purchases, which indicate our in-store purchases. I’ve then mapped the field names for customer transaction history, then transaction ID, then date, the transaction amount, and the product ID. It then populates the next set of fields for customer data. I’ll keep those and move on. I can add additional data if I want. And here I’ve added online purchases to combine them with our in-store purchases. And next I can choose the update schedule, which is the frequency the AI model runs to retrain itself as data is updated with more activities. I’ll leave the default to monthly and hit next. Now I can review everything I’ve configured and start running the model.
- Okay, so what happens then after the model is run?
- Well at that point, we’ll get an attribute-based segment suggestions in the segments tab. The highest percentages will correspond with the highest CLV scores. And these are our most valuable Contoso Coffee customers. To build the combined segment with CLV in churn, I’ll click create segment. Now I want to add the churn segment that I showed you earlier to the CLV segment that I’m currently editing. I’ll search for churn and add it. I’ll keep this keyed off the customer ID and set it to all records. And then I’ll change the relationship to an intersection. What this does is generate all matches where a customer is both high value and at risk of churn. Now I’ll hit save, and that’ll normally take some time to calculate completely. So to save time, I’ll go to one that I previously created, and it’s already loaded here, called High Churn — High CLV Customers. And when I click into this, you’ll see the results of our combined segment. This can be downloaded as a CSV file, but for our marketing team at Contoso, using Dynamics 365 marketing, it’ll be ready for them to use as an audience segment. And pay attention to Claudia Mazzanti, because we’ll be highlighting her journey.
- Okay, so does this then get shown in the customer record, maybe if the customer calls us or visits a retail location?
- Yes, it can. This is the profile from one of our best customers, Claudia Mazzanti, as evidenced here by her lifetime value. In this case, we’ve already customized our customer view to display tiles for lifetime value and churn risk. We can see she’s at a high risk of churn. 0.98 is a high score and anything close to one or above one in this case is bad. We can see her recent activities and she isn’t purchasing as frequently as she used to. And she hasn’t been in our stores in a while or purchased online from us since last November.
- Right, and Claudia is one of many customers then on the high CLV segment list, that’s also at risk of churn. So how would you turn this around?
- So as an analyst, you’ve done your job and you’ve surfaced the risk. And this segment list is immediately available in Dynamics 365 Marketing, which makes it a seamless handover to your marketing team to now take action. Now, as a member of the marketing team, in Dynamics 365 marketing we can see the segment we just created, and we can use this list to create a promotional campaign that targets our 245 high-risk customers. Here’s what our marketing team is proposing. They’ve come up with a series of offers to bring customers like Claudia, who haven’t purchased from us in a while, back to our stores. We’ll build an incentive program that uses either email or text based on their contact preferences and offer this segment of customers a free coffee during their next in-store purchase. We’ll also sweeten the deal. If they refer a friend, we’ll give them a second free coffee and the friend gets one too. Let me show you how to do this. I’ll select the segment and click on new journey. I’ll give it a name, in this case, returning customer. And I’ll keep this as a segment-based journey and leave the defaults in the other fields and choose a start date, then I’ll hit create. Now, as you can see, this created a journey for our segment. A journey is an automated workflow that allows us to define steps that we want in the customer experience. And now we can build this out. When I hit the plus sign, you’ll see all of my options. Here, you can communicate with customers across channels, like email, text, and push notification, and even leverage AI to select the best channel for each customer. You can also respond to customer actions, easily test messages or channels against each other, and even set custom timing for when a message is triggered. Also, when I create a journey, I can set goals so that I can later measure success. So I’ll click into the goal details. Let’s set the goal to drive a purchase. And we’ll say that the goal is met when a purchase is complete, and then we’ll set a threshold for the amount of people needed for this goal at 10%.
- Okay, so then you would continue this by adding and defining different steps in this customer journey, right?
- That’s right; and to save time, I’ve already built outmost of the journey. You can see I’ve already added quite a few steps. On the lower left, you’ll see our refer a friend offer I described earlier, and then towards the top of the journey, you’ll see I’ve set up an A/B test for different offers. Here, for example, I’ve created two different emails, one with a highly personalized offer that I’ll show you in just a sec and another with a less personalized offer. And here I’m showing you a draft version of the highly personalized email. You can see there are variables to add our customer’s first names, and we even know what their favorite drink is to add a more personal touch. And all of this data is fed to us from Dynamics 365 Customer Insights. Also adding the right images to an email can make it more engaging. Contoso Coffee happens to have a large asset library. And so if I jump back over to our email, I’ll select an image. And here Dynamics 365 Marketing recommends two great options automatically using AI. This works because images are auto tagged when you add them to the asset library. So it makes it really fast and easy to pick the right image. You’ll see this one matches it perfectly. And so I’ll go ahead and add it. As I create this email, you’re seeing variables for all of the fields, but for Claudia it will look like the view on the right. And we even see her favorite drink is a white chocolate mocha.
- Okay, so that’s the more personalized mail option, but what are you testing this against?
- So for promotion B, we’ve also created an email with some personalization, but it doesn’t explicitly name the favorite drink. Instead we offer any drink up to a $5 value.
- Okay, so how else then can this help a marketer build out an even more personalized experience?
- Well, we’re still in the journey. So let me show you just how easy it is to add new steps. In fact, I’ll add a step for channel optimization. This helps determine the best way to communicate with our customers. First, I’ll add a step to see if email performs better, and then I’ll add an option for push notification to see if that’s the best way to reach our customers in this segment. Finally, I’ll hit optimize, and this will determine the most effective channel for communication with our existing high-CLV, high-churn customers.
- Okay, and you also mentioned a refer a friend incentive, and since referred customers kind of by definition are new. How do we determine the best way to communicate with them?
- Well, here we want to build a separate journey for our brand new referred customers. Again, I’ve already started a journey to save some time. It has a free coffee offer, but what we want to do is something different, in this case. I’ve added an event which triggers in real time to activate an offer and alert the customer immediately once they connect to our in-store wifi. The customer will receive a message that they are eligible for a discount. And we can use either text messaging or push notifications from our Contoso Coffee app. And this will leverage the built-in AI-optimized channel selection, which runs experiments to determine the best way to reach the customer. And now I can go ahead and hit publish on both journeys and this program will run.
- Okay, so now we’ve gone from the data analyst discovering the risk, then the marketing team also building and executing offers to retain high value customers, all in one connected environment. So how can we see then if our campaign worked?
- Well, once the journeys are running, everything gets measured against the goals you’ve set and all of the contributing activities from your customers. So let’s fast forward a few weeks. We can see the we exceeded our goal of 10% and we’re actually already at 11% and there have been 27 purchases. And on the left here, the thickness of the lines on the journey are proportionate to their success. This is also known as Sankey visualization. And when I click in on the right, we can see the results of the A/B email test. People were more engaged with the offer that named their favorite drink versus choosing a drink of their choice with a dollar value. We can also see the results of our second journey with the in-store promotion that was triggered when a customer connected to our wifi. Notice too, on the top right of the screen, the AI optimized how customers in the store were notified, using either text or app push notifications. And this led to a 29% increase in our goal attainment. So all of this measurement and tracking is automated once you’ve published your journey, and you can continue to check back in on your results.
- And since we’ve fast forwarded in time, going back to our data analyst role though, can they see the impact as well?
- Yes, they can see the direct impact on churn and the new customer referrals. If we go back and look at Claudia, we can see that she’s now at less risk of churn, at 0.21 versus 0.98 from before. And she’s also referred several customers to our business.
- What I like about all of this is that the information flows within different contexts, whether you’re the analyst or the marketer, and because there’s a data management backend, all that integration is taken care of for you.
- Yeah, and although we spoke a lot about journey orchestrations, personalizing customer experiences extends far beyond offers. There’s more you can do around event planning and management, creating and running online surveys and other customer engagement activities.
- And given how much things have really changed in the past year and the way people buy and consume, Dynamics 365 with Customer Insights is a great way to keep a pulse on your customer base, so that you can take the right measures to keep them. So where can people go then to learn more?
- So, to learn more about implementing the personalized customer journeys that I showed you today, you can head over to aka.ms/D365MarketingHowTo, and also a key part of instrumenting this experience is Customer Insights. So you can check out our Mechanics show on how to set this up at aka.ms/CIMechanics.
- Thanks Thiale, and of course, keep checking back to Microsoft Mechanics for the latest tech updates, and subscribe to our channel if you haven’t already. And thank you so much for watching.