When Tony LoFrumento joined Morgan Stanley in 2001, he was given a wide-ranging goal: to create a CRM and business intelligence (BI) infrastructure that would enable the financial services company to transition from a product-focused entity into a client-centric business. In developing a CRM strategy for Morgan Stanley, LoFrumento opted to put his resources into customer analytics — building a centralized data mart, developing predictive models for such areas as client profitability and performance measurement, and bringing in BI, analytics and Web-based reporting tools to analyze and disseminate information — rather than deploy an operational CRM platform.
“My theory was, ‘Let’s get our analytical house in order prior to venturing into an operational CRM platform,” says LoFrumento, executive director of business intelligence and CRM for the New York-based company’s Individual Investor Group, whose retail operations have more than 450 branches and some 3 million customers. “Operational CRM systems can be working perfectly, but if they don’t have the business intelligence backing them up, you’re building some pretty expensive Rolodexes,” he says.
Morgan Stanley’s decision spotlights an ongoing debate as organizations move to customer-centric technologies. With limited budgets, companies often have a choice to make. They can spend on operational CRM applications, which automate processes in sales forces, marketing departments and contact centres, and enable them to gather and share customer information. Or they can adopt a BI strategy toward CRM, putting analytics applications in place to leverage data sources they already have, and perhaps move to packaged operational applications later. Views vary, but many believe the real return on investment lies in leveraging both — particularly in large, business-to-consumer organizations.
Laura Preslan, an analyst at Boston-based AMR Research Inc., says most BI technologies being applied to CRM are analysis and reporting tools, which are typically used to review historical data and validate decisions. The future, she says, lies in leveraging predictive analytics, which model various business scenarios, and lay the groundwork for effective change. By coupling analysis and predictive capabilities with operational systems, companies can, for example, model customer behavior to develop targeted marketing, determine customer profitability to establish and deliver appropriate service levels, and tell customer-facing personnel what products or services to offer.
Analytics may represent the next phase of CRM, but the companies that have invested significant resources in the operational phase aren’t eager to spend more. “Customers really rebel when they’ve spent US$5 million on an operational CRM package and now have to spend another US$1 million on the analytics,” says Steve Bonadio, an analyst at Stamford, Conn.-based Meta Group Inc.
One reason CRM systems fail to meet expectations is that businesses don’t know how to measure their progress or further improve the new processes, says Bonadio. “That’s where analytics comes into play. Operational and analytical CRM are highly complementary — you can’t do one without the other,” he says.
Much of the negative publicity about CRM comes from companies spending large amounts of money on collecting operational data that they haven’t applied to decision-making, says Preslan. “The good news is that these companies now have the data, but it’s a tough job getting the right tools and business processes in place to get it back out,” she says. Difficulties include centralizing what may be a plethora of siloed databases, creating a single view of individual customers across multiple channels, and implementing a closed-loop, embedded analytics approach wherein operational systems feed analytics and the resulting changes have a positive impact on operations.
Morgan Stanley has already seen benefits from creating predictive models for certain segments of its customer base, enabling the company to refine its marketing efforts. “Think about all that money you’re saving not going after customers that the model says will have a very low response rate,” says LoFrumento. “You can run a marketing campaign based on a predictive model on propensity to purchase, which is much different from how you ran it in the past when (campaigns) were likely based on basic audience selection.” Morgan Stanley uses marketing automation and campaign management analytics applications from SAS Institute Inc. in Cary, N.C., and San Jose-based Business Objects SA OLAP and Web-based reporting tools, which allow it to package reports and distribute them throughout the organization.
Management company AGF Management Ltd., which is in the midst of a CRM deployment that involves both operational and analytical systems, likewise expects to see significant business improvements from predictive modeling. Toronto-based AGF’s goal is to become a preferred supplier to its intermediary distribution force — the financial planners, stock brokers and insurance agents who sell AGF products to investors.
The company has already implemented PeopleSoft Inc.’s service module and is in the process of deploying its sales and marketing modules. For analytics, it will use the PeopleSoft’s integrated analytics tool, as well as tools from Toronto-based Angoss Software Corp., a PeopleSoft partner.
“We just started working with PeopleSoft to create an enterprise-level system to support our overall CRM requirements. You have to build that operational capability and pipeline and begin that entire cultural change within the organization to prepare the way,” says Stephen Elioff, vice-president and CRM program director at AGF. “But we believe that analytics are going to drive the ROI for the project. Analytics are going to help us more effectively recognize behavioral purchase trends and put us in a position to more rapidly respond with the appropriate level of resources to the opportunities in front of us.”
At Center Parcs Ltd., which operates vacation parks throughout Europe, deployment of predictive modeling tools from Netherlands-based DataDistilleries has brought quick improvement in direct marketing campaigns, says CIO Richard Verhoeff. Though they’re not yet tightly coupled, the Windows NT-based modeling application and the company’s operational campaign management system, from Suresnes, France-based AIMS-Software, share the same DB/2 database. Center Parcs also uses analytics tools from Chicago-based SPSS Inc., which acquired DataDistilleries last year.
“(DataDistilleries) really brings data mining into our production process,” says Verhoeff. One immediate payback was the ability to develop promotions for two difficult-to-sell December holidays by allowing marketers to recognize a specific population that likes to travel during that time.
Achieving the ROI that executives anticipate from analytical CRM requires significant preparation, including centralizing data for analysis, creating a holistic view of each customer and linking systems to take advantage of findings.
Morgan Stanley has created a centralized data mart, which pulls data from all its mainframe-based data silos. It performs an extract, transform and load process monthly to extract data, and uses a sophisticated householding algorithm that groups all data by address to create an individual customer ID. “If you don’t have accurate client profitability and lifetime-value calculation capability, how can you apply the appropriate level of service and offerings to each client? Most companies can’t, and then their top clients are subsidizing the less profitable clients, which means they’re more vulnerable to be cherry-picked by competitors,” says LoFrumento.
Though systems integration is a priority in many analytical CRM projects, companies “shouldn’t be as concerned about integrating systems through systems interfaces as about getting the data into the single data warehouse from which you’re drawing to do your analysis,” says AMR’s Preslan. “That’s a huge effort because that’s basically an enterprise-wide project. But without that you have each analytic application with its own little data mart and you end up getting 80 different reports each month.”
Preslan says IT departments can start an analytics effort by interfacing certain data marts, but they’ll still have to review the corporate data dictionary to determine differences in how each system defines various attributes, such as customer IDs, so it makes sense to move toward a centralized data architecture. Eventually, says Meta’s Bonadio, companies will benefit from advancing to a closed-loop architecture to exploit the benefits of both their analytical and operational platforms. This comes from a workflow architecture that incorporates event management principles to trigger appropriate actions. “After you’ve developed KPIs (key performance indicators) and you’re able to set thresholds based on conditions or rules, they’re able to trigger some action, generally in an operational system, that says something needs to be changed,” says Bonadio. Also important is data delivery to personnel through various user interfaces, such as portals, where operational and analytical information can be displayed side by side and in context.
Morgan Stanley will deal with such issues when it implements an operational platform, which it eventually plans to do via a contact management system. “Given the pressure our industry was under the past few years, (operational) wasn’t at the top of our agenda, but if this effort is going to hit on all cylinders, the operational side is needed,” LoFrumento says. However, he adds, “if you’re starting from scratch, do the analytical first, because there’s immediate payback to all that knowledge and you’re perfectly positioned when you do roll out operational. Furthermore, analytics is not as expensive, and there’s immediate payback because it provides the information you really need to run the business.”