AI Will Propel the New Customer 360



he abundant growth of data, maturation of machine algorithms, and future regulatory compliance demands from the European Union's General Data Protection Regulation (GDPR) will shift the landscape for creating a single source of the truth for customer data. Historically, companies used combinations of CRM systems, master data management (MDM), and data lakes to create that one source, but all have failed to live up to the expectations of front-line business users in areas such as marketing, customer care, and digital commerce.

Looking ahead, however, it is clear that the new requirement will be investment in customer intelligence platforms (CIPs) that do more than consolidate a single view of the customer: they add a layer of data governance, synthesis, and identity, which powers a dynamic customer graph to fulfill the vision of contextual experiences.

Improving customer experience demands an approach that takes into account all of the tools, processes, and data across the customer journey. This complex process usually involves dynamically maintaining a single source of truth about each customer to drive personalized experiences based on individual preferences and behaviors. Businesses to date have primarily invested in systems of record, such as legacy CRM and ERP, to serve this purpose. While these systems are critical for managing internal operational processes, they are typically not effective for consolidating customer information at the pace of business change today.

Structured data from operational data stores now provides only a small slice of the overall data needed to improve customer experience. IT departments previously invested in MDM and data warehousing technologies to consolidate information associated with customer profiles. The emergence of additional unstructured data, however, further relegated traditional CRM, MDM, and other systems to just another silo.

Businesses today need to incorporate the exploding growth of unstructured data from IoT sensors, social data, behavioral data, location data, and even third-party data to truly have a single source of the truth. In response, digital marketers and agencies have adopted data management platforms (DMPs) enabling companies to target campaigns to anonymous audiences across third-party ad networks and exchanges. A few DMPs have evolved into customer data platforms (CDPs), aimed at creating unified profiles of customers from multiple sources of data regarding both known and unknown individuals.

As the universe of what is knowable about customers is expanding, new machine learning technologies evolve to help businesses see further and deeper, improving business decision-making. Combining human expertise with machine intelligence can be a powerful tool, since human interpretation alone can miss contextual clues in large data sets. According to data from 451 Research's "Voice of the Connected User Landscape (VoCUL): 1H 2017 Corporate Mobility and Digital Transformation" study, 82 percent of businesses say that ML for automated contextual recommendations is important to creating personalized customer experiences.

Enter the next-generation CDP, which 451 Research defines as customer intelligence platforms (illustrated in Figure 1) that drive contextual experiences. The advancements in predictive ML intelligence build on a variety of algorithms to achieve real-time one-to-one capability (ideally in fewer than 20 milliseconds). CIPs are not just about the data, but also the potential for delivery of dynamic rich media content, including images, videos and voice.

Figure 1

CIP

A CIP must go a step further than a CDP by synthesizing data that dynamically links customer to customer and data to customers using an optimized mix of matching techniques. It provides context from raw data for relationship discovery, with graphs, columnar data stores, and in-memory high-performance indexes to drive multiple versions of the truth for different use cases. As it ingests and synthesizes more data into the customer 360, a CIP platform must also become more intelligent in identifying important trends and information for each customer and better at summarizing the important intelligence for specific business users. Synthesis and reasoning must work in balance to ensure the CIP is usable; as more data is synthesized and the customer 360 becomes deeper and richer, the CIP must get better at summarizing the important intelligence for specific business users.

Automated reasoning helps to make inferences and enrichments on each customer profile and also helps line-of-business users predict customers' future actions, such as churn, propensity to buy, proximity, and location, etc. It provides a deeper understanding of individual customer journeys and unique interactions, combined with transactions, to accurately understand and improve customer experience.

But in the end, the bottom line is that companies must do the following:

  • View data lakes as IT projects. While they do offer benefits, many data lakes aren't architected for real-time analysis, have limited APIs, and potentially fail to perform proper customer matching. They have also become corporate warehouses for storing very specific data sets (HIPAA, credit card info, etc.). However, organizations can leverage their general-purpose data lakes into customer data lakes, using CIP technology to create the customer 360 for deeper analysis and insights.
  • Continue to adapt single-source-of-truth customer data strategies for non-technical marketing and customer service. Lack of skilled users and adequate data and content are already huge barriers to effective customer-facing use cases; successful tools must be useful to those who aren't data scientists. CIPs must also play nicely with line-of-business tools by providing valuable information to improve the sales, service, or marketing applications that employees already use.
  • Understand the alphabet soup differentiation. MDM, DMP, CRM, CDP, and CIP platforms all offer a variety of benefits, and businesses are still searching for the holy grail solution. It's important to remember that success means empowering the end-user use cases, not just the approach of he who holds the most data wins.

Sheryl Kingstone is research director for business applications at 451 Research.