Commentary--Is managing customer relationships really that difficult? Customers expect you to know who they are, know the nature of their existing relationship with you, know what you told them last week, fulfill the product or service you promised them, and treat them fairly and consistently. Simple, right? With the proliferation of customer interaction channels and the number of product and service bundles, recognizing who customers are and what you've told them is hard enough. Now let’s add in the complexity of fair and consistent treatment. How do you ensure that customers who received a 10 percent discount via direct mail get the same treatment when they research the product on your Web site? And how do you ensure that the way you treat your customers is also profitable for your company?
Enterprises that are driven by the need to more profitably manage customer interactions--and to improve the customer experience--are starting to engage in projects that drive deeper customer insights into each interaction across channels. As companies look to integrate analytics into front and back office systems, important architectural and application-specific considerations emerge. Not surprisingly, service oriented architectures are well suited to this challenge.
Customer experience programs typically start by identifying an organization’s high-level strategic customer segments and its key strategies and messaging for each segment (grow, retain, sustain, etc). This is the traditional starting point for direct (or database) marketing. It sounds simple, but as these strategies are implemented companies usually have to consider competing operational constraints coupled with customer preferences, permissions and propensities. Moreover, the effectiveness of these strategies should be measurable, and measurability requires testing, response analysis and reporting.
For example, what starts as a simple rule to check a person’s credit rating to determine platinum card eligibility can quickly evolve into calculations of lifetime value, credit limit, etc. Then, you’re configuring automated rules for credit limit increases and special loyalty promotions with conditional rewards based on account activity within specific time periods. Before you know it this seemingly simple set of rules has quickly expanded into a full blown enterprise application that must interoperate with many existing operational systems in several distinctly different operating modes--namely batch, real-time and event-triggered scenarios. And since these projects usually start with one particular customer interaction point like the Web or contact center, the temptation is to code these rules directly into that particular operational system or license an add-on product. The danger with this approach is that customer data and rule replication across operational systems dramatically increase project risks, decrease solution agility and promote the silo problem we set out to solve initially. (Remember, we are trying to improve a customer’s experience across channels by delivering seamless, relevant and consistent treatments.)
These problems are compounded by the need to analyze the effectiveness of customer treatment strategies across channels. Cross-channel performance analysis requires the ability to define relevant response levels and types across multiple channels and to properly attribute the effectiveness of each treatment strategy. If an offer was refused in the call center, did an individual go on-line two days later and purchase the service after all? And how many customers were going to purchase or renew the service anyway? In order to measure the incremental benefit of a customer treatment you need to incorporate statistical testing and control group methodologies. This cross-channel response analysis is critical to meeting our business requirement to deliver profitable customer experiences and enable objective measurement and assessment of ever evolving marketing strategies.
The good news is that these capabilities already exist within modern enterprise marketing platforms--it’s traditionally called campaign management. Done right, campaign management is that elusive killer app that makes analytics actionable at the right time and place across your organization by leveraging the information in your existing customer marts or enterprise warehouse.
The "marketing brain"
Many successful IT and marketing organizations are moving toward what some industry insiders have begun calling a "central nervous system"--a system that collects and filters data across channels, leveraging a centralized “marketing brain” to analyze and determine the correct treatment strategy for any given situation. This nervous system is enabled by service oriented architectures, which allow marketers to plug emerging marketing functions into existing operational CRM systems and enterprise solution frameworks for end-to-end, consistent automation of customer treatments.
The evolution of the central marketing nervous system makes sense for several reasons. Not only do marketing organizations normally dictate customer treatment strategies, but they also own the data infrastructure and core analytical methods that form the basis for implementing customer data-driven treatment strategies. They have spent years creating, cleansing and analyzing information from operational systems into an enterprise-wide view of customers and their behavior. They also have the resources necessary to perform the deep analytics needed to determine the most profitable strategies for any particular interaction and to analyze the results. Finally, many have also invested in sophisticated campaign management tools that are well suited to serve as the “marketing brain."
A good campaign management system will allow the enterprise to look across disparate data sources to determine the best strategy for each customer. These systems are also designed to operate across interaction channels. Leading edge systems also include a common offer management repository that is a critical component to enable coding, tracking and analysis of customer activity across channels, time and products.
Let’s think through how the “platinum credit card” scenario mentioned earlier might work in a services oriented architecture driven by a centralized “marketing brain”. Each customer interaction point simply posts a message with the relevant information--customer-ID and contextual information (such as a customer’s interest in a platinum card)--to a central decisioning service.
This service performs the necessary back-end credit checks, scores the customer for expected lifetime value and looks at the available platinum card offers to determine which is most appropriate for the customer.
Once the offer is presented to the customer, it is recorded as an offer in marketing’s contact history data store. The decisioning engine then monitors customer activity across multiple transactional systems and data stores to determine if there was a positive response or a related product activity that can be attributed to the initial contact or offer. This approach can give marketers (and businesses) a real-time view of offer performance across channels, as well as the ROI of specific treatment strategies, allowing them to dynamically refine and manage treatment strategies.
Enterprises embarking on projects to enhance the customer experience and ensure consistent value-based treatment strategies should take a hard look at technologies they already have in-house. Today’s leading marketing automation tools provide a strong foundation upon which to build a centralized, service-oriented decisioning architecture that has a faster time to market and lower total cost of ownership than distributed channel-centric approaches.
biography
Todd Piett is a senior segment manager at Unica Corporation. He is responsible for new product initiatives and ensuring customer requirements are incorporated into the product development and marketing of Unica's Affinium software.
