Type in your prompt above or try one of these suggestions
To thrive, businesses today must rely on data to better understand customers and personalize new products, services and experiences based on the information that data provides as a reusable and transferable asset. Data allows a business to better target market activities across channels and share actionable intelligence throughout the enterprise. The challenge is that many lack the tools necessary to integrate online and offline customer data rapidly enough to generate insight before it becomes irrelevant. Perhaps this is why estimates show the global Customer Data Platform market size growing from $903.7 million in 2018 to $3,265.4 million by 2023—at a CAGR of 29.3%.[1]
Customer data platforms (CDPs) collect and store information related to customers as unified data to reveal deeper insights and intelligence quickly enough to surpass broad-scale marketing. Beyond enabling an enterprise to thrive in an omnichannel and privacy-conscious marketing world, CDPs represent exactly what enterprises need to truly transform their businesses: a way to proliferate customer-centricity throughout their business. However, there are a few things they need to get right:
Data is the key to knowing your customer. It creates the 360-degree view that allows a business to understand its customers’ intent, preferences and needs. Every customer interaction with every touchpoint has the ability to generate—and be informed by—customer intent.
The good news is that actionable data is available in abundance. The bad news is that these pearls are usually hiding in a sea of noise. Businesses must decide what to collect. Whether the data comes from media usage data, call center interactions, mobile app clickstreams, transactions (purchases), in-store or cross-channel interactions between brand and customer—it’s all valuable and can be a business differentiator. The successful CDP initiatives orient collection around a strong Identity Strategy, where information about customers is collected according to a definition of signal. Signal in this context has four parts: A. the customer’s intent at the moment of collection, B. their expectation of the brand (and their experience with it), C. the use-case for that data to drive business outcomes and D. the value of the synthesis of other data with the data that is collected. Signal can mean different things to different businesses - while the components are always the same, the definition is as unique as the business itself.
In addition to determining what signal means and therefore what to collect, every company must create the means by which to collect it. These solutions combine aspects of use experience and technical design skills. Should some data be collected in real time as a stream? Which? What should we define as a customer’s primary identity as opposed to secondary? Do we actually need a new experience to be able to collect the right data (and if so, how should we provide value to the user so that that data flows in a way that respects their intent? Should stitching be done in batch mode so that end-to-end data collection occurs every night? These questions should be answered in the CDP’s input/output infrastructure.
The information stored in data has a higher potential of more value when combined with other data. CDPs offer a platform for stitching together data to create a customer “graph” or user profile. There are so many potential digital identities for each person, a proper CDP can connect the dots to create one digital identifier for a customer. Without this critical function, a business is left guessing. Is this the same customer or not? Did they just change their address? Is this a bot or fraudulent account?
So how will you stitch it? This is where it gets a bit technical. For every use case, the business will have to determine if it needs deterministic data (when two pieces of data often have the same identifier in common, making induction possible) or probabilistic data (when we have to rely on a probability standard and use deduction to identify the user). Regardless of which identity resolution method is applied to which data source, data pipelines use one or the other to stitch data from first-party, second-party, and third-party sources to output a view of the customer that can meet reporting, analysis and activation use cases.
Data must be available, understandable and actionable so that the business can develop use cases for how to monetize data across marketing, sales, e-commerce and supply chain. Creating an artificial intelligence/machine learning model behind each use case will reveal opportunities.
Often, it’s language itself that prevents this sharing from occurring naturally. Just as systems need to interpret coding languages to understand how to process data, every one of us speak and understand each other for data to be truly usable across the enterprise. For instance, a marketer may say, “I need to understand customer behavior and what content a customer looked at” and the IT person may say, “I want to do a select of customer information column plus the clickstream column.” Both may be talking about the same outcome, but in different parlance. A term you hear more and more as part of data science teams is “Data Translator” or “Data Strategist,” for this reason.
Through a properly built CDP, a business can create a central intelligence hub by which to discover and use customer information, almost like a catalog of usable data. This hub ideally puts a user-friendly wrapper around the technology interface, creating a common language so that users of all types have the data literacy to access what they need in the format they need it. Often, the full set of use cases to apply the synthetic value of unified customer data grow over time, and once they are combined they grow quicker. Getting your CDP right can make or break a business transformation effort; without one, enterprise scale customer-centricity is fundamentally impossible.
Read the rest of the series:
Related Articles
