The sixth scenario or business use case for Customer Data Platform (CDP) technology is Realtime Behavioral Analysis. It encompasses a potentially tricky set of services and brings some important challenges around performance and security, so if this interests you, you'll want to investigate closely.
Based on Machine Learning...But Real Time
Most CDPs support some level of machine learning. What this means is that they have machine learning algorithms that are trained on vast amounts of data and are then used to make a prediction (e.g., what should be the next best action in a marketing campaign). This is of course simplified version but you get the idea.
Typically machine learning works on historical data. The algorithms use existing data to train a model that then predicts a future outcome. This data is often static, meaning it remains more or less same for every prediction. Sure, you might refresh it at periodical intervals, but then it remains same for every prediction, until you refresh it again.
Now, while this is good enough for many use cases, sometimes you need more. Imagine a very high-traffic site in which you would like to personalize your offers based on realtime visitor behavior. What this would mean is that you will need to take that streaming user data, make it a part of your training set, train your model on that data set and then come up with a prediction — in real-time. This capability is often referred to as online (machine) learning.
How Do the Tools Fare?
Most CDP vendors will claim they have machine learning capabilities. But very few can apply machine learning to streaming data in real time. CDPs that excel in this scenario offer more advanced algorithms that can work against existing data as well as live streaming data.
RSG's research can help you with your analysis. If you subscribe, you can use our RealQuadrant Shortlist Generator to find out which CDP vendors excel at this scenario. RSG's evaluations also call out specific capabilities for this scenario for each CDP vendor.