Recently NGDATA was the CDP Institute’s guest for a webinar on how to get the most out of your Customer Data Platform (CDP). How do you pick the right one? And what do you need to consider once you have chosen a CDP?
In the other article of this webinar blog series, we already saw how to get started with a CDP and what to consider when picking your first use case. When you got that covered, what’s next? Your program should not end after the first use case implementation. After building momentum, it’s time to build out more impactful scenarios.
So what next?
Once you have determined your first use case, how do you pick it up from there? After deploying a few initial use cases that return immediate value, you’ll want to add longer-lead projects that create higher long-term value. These will take longer to complete but have the potential for greater impact. This is linked to increasing precision when engaging with your customers. With predictive model scoring, you provide your customers the most relevant experience. Reach them through the right channel with omnichannel orchestration. And finally, as timeliness is essential, real-time responses to a customer request or behavioral changes demonstrate how truly customer-centric you are.
For specific use cases, the CDP incorporates data from other sources to execute predictive model scoring in real-time. To calculate a customer’s propensity score, you can have simple formulas such as someone’s income per year in relation to their age, but you can also use actual predictive models that are trained on data. To define a customer’s propensity to get a car loan, various parameters will be able to signal whether you’re dealing with a high or low propensity to convert. In our CDP, we call these parameters ‘metrics’. So for predictive scoring, you will use a model. If you successfully want to implement real-time interaction management, and if one of the metrics is the time that has passed since the prospect last visited the website page of car loans, you need to have real-time access to all the necessary data, for instance when someone just visited the page. Besides that, you also need to be able to score the model of all metrics, including the time since the last page visit, in order to offer personalized interactions with the customers that have the highest propensity.
The next thing to tackle besides adding data sources is to make sure everyone in the organization is on the same page. Get all departments aligned as the people from IT services or sales would also get great value from a CDP, but it generally is not where you’re going to start because those departments have their own processes.
And finally, go omni-channel. Integrating one channel can be challenging, but coordinating across channels is even trickier as the orchestration angle adds another level of complexity. Coordinating interactions with each customer across all channels can extend beyond marketing to all departments and throughout the entire customer lifecycle. In addition to data and channel integration, this requires complex analytics and personalization rules that take into account elements such as historical behavior, current context, channel preferences, and available offers. And in order to optimize system decisions over time, it’s important to capture results.
Focus on the key components for a particular use case, and really unravel the sources necessary to dive into that use case’s KPIs and building blocks.
If credit card utilization is your first use case, for instance, you can start by following the transactions and building an individual customer behavior profile to notice when transactions are increasing or declining, and then reach out to that customer in the mobile app to provide them a custom offer. Next, you want to increase the customer-centricity, so eventually, you would start to move the needle on those longer-term metrics such as NPS, customer satisfaction, and retention. So airlines, for instance, can start with flight bookings and add information like check-in and checkout, and on-board services data. Telco & media could start by checking the usage of telephone services and add data services. Or start with product ownership like digital tv, mobile data, and expand to tv content packages. So as you expand the data coming in and you expand your ways to interact with the customer in accordance with that data, you start to move along those long-term metrics as you move forward.
Focus on the key components for a particular use case, and really unravel the sources necessary to dive into that use case’s KPIs and building blocks. And while you want to start small and build up from there, you also want to make sure you focus on gathering all technical requirements upfront based upon that use case. That helps streamline the entire process because organizations typically go to an IT source for a particular data set requested by that department, but those might not be in line with what your use case requires.
Real-time approach with a CDP
As customer expectations have evolved to immediate relevancy, it puts data-driven in a different perspective when it comes to time. Real-time interactions are technically more challenging as these require tight, high-speed integration between the CDP and customer-facing channel systems including web sites, mobile apps, and call centers. You need to have the right response time and have all the touchpoint systems wired up to go back and forth with the CDP in real-time.
When talking about a real-time approach with your CDP, we can distinguish three types:
- Real-time data ingestion. In this case, you want to be able to see within a second in my CDP someone’s website behavior so you can, for instance, share that info with the call center while they’re on the phone with that customer.
- Real-time profile access. This is more demanding as you want to look up a customer’s data in the CDP when they visit the website. This means the website has to call into the CDP and then get back within – in this case – a fraction of a second with that personal info so you can run that through the personalization engine and orchestrate the right offers.
- Real-time recommendation. The CDP receives a request from the website, then runs processes for predictive modeling scoring and creates a recommendation within the CDP. The outcome is returned by the CDP to the website’s frontend. So now you’re not just returning data, you’re actually doing some processing, scoring, recommending, and returning the results of that process.
Real-time is very powerful to support your customer by immediately responding to his needs, yet it has specific requirements when it comes to the data elements you need and how quickly you need them.
So while real-time data is very important and having instant access to it for the orchestration, using the info in an appropriate way to service your customers should always be front and center.
Real-time reality checks
Understanding what should be real-time, near-time, or maybe just batch, really helps you prioritize your efforts. For example, spending a lot of time trying to enable segmentation data may not be meaningful in real-time because those attributes might have lagging update times, however, ensuring that web behaviors are flowing in real-time can enable a fluid conversation with the customer before they leave their web session.
More commonly, channel data that captures what your customers are doing, is really key to that real-time orchestration as it lets you act upon those actions at the exact same time. If you want to effectively execute an offer to your customer, you need to understand what they’re doing at that moment and not a week later. However, with that being said, understanding your customers and reacting or orchestrating in real-time are often two very different things. In the case of a hotel, for instance, when you send your guest a push notification to his app while he’s checking in, offering him a free coffee, then you bet, he wants to receive that in real-time. But when a telecom customer is on the phone with an agent to upgrade his internet package, it might not be as meaningful if he immediately receives an email to inform about the higher cost and repeating the info the agent already provided. He may prefer to get that email later in the day.
So while real-time data is very important and having instant access to it for the orchestration, using the info in an appropriate way to service your customers should always be front and center.
Conclusion
In the other part of this blog series, ‘The first use case’, you could already read how to get started and how to determine your first use cases. Our advice there was to start simple and slow. After this, you can pick it up a notch by adding longer-lead projects that will create higher long-term value. These projects are usually more complex in terms of technical and organizational efforts, they will typically take more time to complete, but have the potential for greater impact.
Want to see how real-time interactions come to life?
