There is ample confusion in the marketplace about customer data platforms (CDPs) and digital transformation in general. With traditional tag management vendors identifying as CDPs and cloud vendors pushing into the market, the ability to understand differentiation between solutions is complicated. The Redpoint product marketing team recently had a discussion with an analyst who specializes in personalization and customer experience (CX) best practices about Redpoint rg1. During this discussion, the analyst identified that a key difference of Redpoint’s approach is its ability to serve as a high-performance engine when it comes to data management, automated machine learning and intelligent orchestration. Regardless of the endpoint, Redpoint could serve to accelerate digital transformation in enterprises in an array of industries.
With a consensus that data management is the strength of the platform – encompassing everything from orchestration to real-time decisioning, breadth, depth, quality, identity resolution and latency – the discussion then centered on three key elements for how rg1 stands out from the crowd in the CDP market: driving prescriptive, predictive and insight marketing use cases.
Take Prescriptive Capabilities to a New Level with AML
According to the analyst, the Redpoint rg1 automated machine learning solution illustrates the prescriptive capabilities. In a use case such as a retailer using automated machine learning (AML) in a dynamic pricing campaign, the prescriptive element showcases the vast divide between manually setting an arbitrary segment (arbitrary by age cut-off, geolocation, income, etc.) and using AML to let data determine the optimal segments and sub-segments.
For any retailer that has tested dynamic pricing with an ultimate goal of finding a price that’s fair to the consumer while also maximizing revenue, there’s always a balance between customer needs and company needs. For many companies, human beings manually determine where that dividing line is – setting dynamic pricing rates by age and gender, income level, etc. But the dividing lines tend to be arbitrary. Why, for example, would a 50-and-over male be the cutoff vs. 45-and-over? Or 46?
Prescriptive software capabilities – as exemplified by rg1 – ensures that data supports every segment. If a marketer possesses broad and deep data sets – particularly historical data that link purchases to certain behaviors and attributes – it becomes relatively straightforward to let the software determine the optimal dividing lines, based on hard data rather than human intuition.
Predicting Customer Behavior is Just a Starting Point
Marketers also care deeply about predictive capabilities – for use cases such as determining a likely outcome based on variables such as which image or content to use, what time of day or day of the week to send an email, email frequency, what content to send on a mobile device vs. email, etc.
Predictions, in this sense, extend beyond predicting customer behaviors to include predicting interactions between customer behaviors and other aspects of the system (which product, what channel, etc.). Really, to include any variable in terms of how a marketer engages with a customer.
With rg1, data collection is a foundational requirement, and is not limited to simply measuring customer intent. Instead, every interaction is an opportunity to build understanding, with the marketer collecting data across a set of experiments such as A/B and multivariate testing to validate and optimize outcomes. Experiments using both classical rules and predictive models drive and extend machine learning with more data and customer feedback. Marketers quickly discover which combinations or versions are acceptable, and can test items and situations such as an image, a treatment, time of day, a channel – all at once across a wide set of variations.
Insight: How Successful Was My Campaign, and Why?
Prescriptive and predictive capabilities in a CDP form a good foundation, but they will not answer all the questions pertinent to a marketer, who of course will want to know why a campaign was or was not successful. Why did a prescriptive and/or predictive tool work?
In our analyst discussion, he brought up the idea of “insight” – analyzing and visualizing success factors of a campaign, similar to Business Intelligence tools. To reduce operational gaps, rg1 includes an Insight tool to showcase campaign and interaction responses, both current and historical. In this way, rg1 already aligns with the marketers’ need for better “headlights” rather than just the “rearview mirror” of offline BI tools.
Insights extend to the prescriptive models discussed earlier as well. When rg1 builds a cluster model, it also defines a decision tree for how a customer/prospect falls into a cluster, allowing a marketer to examine what it means to be in one cluster vs. another and the decision that led to that distinction. And the model and Insight panel can match customer behaviors to predictions across time in multiple campaigns and interactions. This capability will test the efficacy of the tool’s predictive and prescriptive might – and more often than not showcase human limitations and the need for predictive and prescriptive tools in the first place.
An Insights dashboard in RPI provides an easy way to drill down into the specific details of a rule, an audience or a piece of content to see how any single insight contributes to the success of a campaign. It provides marketers with an easy way to break down campaign components and analyze one component against another, or against itself across time.
Using Insight together with prescriptive and predictive tools and techniques provides marketers with a clear roadmap for determining the kinds of messages to build and how to present those messages to the right audience that will optimize business goals.
The roadmap will provide answers as to what image, channel, device, time of day and other variables will most resonate with a target audience – laying out a clear path to follow for all of the core tasks a marketer needs to implement a successful campaign.
Not All CDP’s are Digital Transformation Engines
One of the things I was most struck by in this discussion was our shared experiences in conversing with potential CDP customers who recognize the need for a solution but aren’t quite sure about a specific use case for their business.
There seems to be a consensus that meeting customer expectations for a superior customer experience requires some level of personalization, which means no longer putting customers into arbitrary segments. Less understood, however, is that audience segmentation only scratches the surface for prescriptive, predictive and insight capabilities. The marketer needs these tools and a deeply experimental mindset to drill down and discern the potential of an audience. Identifying customers within an audience is one thing. It’s quite another to identify what the best experience is for each customer in an audience.
To do this successfully requires approaching data collection with an open mind. Customer data is not limited to PII and other identifying characteristics. It’s a closed-loop cycle that includes data about the offer that was presented, how it was presented, how the customer engaged with the offer. This approach underscores the need for a solid data foundation as the bare minimum for an enterprise CDP. Prescriptive, predictive and insight tools are key ingredients for a CDP to truly be considered a digital transformation engine.