Intelligence In The Here And NowIntelligence In The Here And Now

It's the moment of truth: Your customer is about to complete a transaction. Will this be the beginning, middle or end of a beautiful customer relationship? Does this transaction — and this customer — fit your organization's strategic vision of how it will prosper in the marketplace?

The pressing need to answer such questions has become a huge driver behind data warehousing and business intelligence (BI). As Jill Dyche discusses in "Get Your Customer in Focus," BI analytics are redefining customer relationship management (CRM); companies are finding that, unless coupled with fruits of timely data analysis, CRM applications fall short of delivering competitive advantage. "Without analytics, how can I say which of our customers is most likely to come back to us?" says Atique Shah, vice president of CRM and technology solutions at Churchill Downs, one of the largest owners of race tracks and off-track betting parlors in the United States. "CRM by itself can't give you the expected return on investment."

Just as important as analytics' contribution to CRM, however, is how CRM is altering the priorities of the supporting data warehousing infrastructure. Strong interest in marketing applications — software purchases are rising 13 percent annually, making it the fastest-growing segment of the CRM market, according to Gartner — is increasing the focus on BI, predictive analytics and other means of applying data smarts to improve the efficiency and effectiveness of marketing campaigns. Conventional historical analysis and the extract, transform and load (ETL) activities essential to gathering and preparing the data are obviously important. But the drive to improve the timeliness and lasting value of marketing applications is becoming a magnet pulling BI, data warehousing and data integration toward actionable, "real-time" intelligence.

Let's look at how IT organizations are addressing key issues of performance, data integration and predictive analytics to support CRM objectives.

Less Latency, More Opportunity

The time lag between market analysis and action in the form of new campaigns or other activities adds up to lost opportunity. Credit marketing organizations even have a measure for campaign latency, called "days not yet in the mail." Reducing the time by a day or two can mean thousands if not millions of dollars, both in savings due to process efficiency and potential revenue gained by reaching customers at just the right time.

Automation is changing the face of marketing, giving organizations the ability not only to coordinate a larger number of campaigns, but also to better allocate resources to the processes most important to customer-centric business goals. Multichannel campaign management is increasingly blurring the distinctions between offline and online channels; automated processes enable more event-driven marketing, where customer behavior triggers relevant offers that are delivered through the most appropriate channel. And, lest we forget, latency is the target of grander, extra-enterprise visions of end-to-end process management. Organizations would like to transmit event information — that is, the customer's response to an offer — to demand-driven manufacturing operations, which then "pull" products and services from suppliers and business partners in real time to fulfill orders.

Therefore, customer data analysis must keep pace with the new marketing machine. Some organizations are using business rules to develop customer intelligence and express that knowledge in a way that process management understands. Faced with a daunting number of account holders and records on file, for example, a large financial services company might use customer data to inform business rules management software, such as Fair Isaac's Blaze Advisor, Ilog's JRules or Pegasystems' PegaRules. The rules engine would manage the automated logic behind the modification of marketing and sales strategies. Without rules and process automation, it could be hugely labor intensive to repeatedly query the data through BI tools to decide on the most profitable business strategy. Along with time and cost savings, the rules would express in process logic the metrics defined by the company to achieve business performance goals. In this way, the business rules help align business goals with the overall execution of CRM strategy.

Speed for the Masses

Customer and market analysis involve both a large volume of queries as well as significant query complexity to gain insight into multiple attributes. The speed with which analytic tools can identify, profile and segment customers often depends on the computing power available. Thus, the cost of reducing latency has been high, usually requiring massive investment in an array of data warehousing software, servers and often a parallel processing infrastructure. Typically, this price tag gives a decisive competitive edge to large organizations with big IT budgets. It's clear that Wal-Mart, for example, owes a good deal of its continuing growth and success to strategic investments in customer intelligence. The company reportedly has more than 450 terabytes of data stored on NCR/Teradata systems.

Epsilon, a leading relationship marketing services firm, chose to gamble on technology from a fairly new kid on the block: Netezza, developer of data warehouse "appliances" that bundle relational database software, storage and server technology into single package. "What if what they're saying is true?" says Mike Coakley, vice president at Epsilon, recalling his team's reaction after its initial look at Netezza Performance Server (NPS). The bundle presents something of a "back-to-the-future" concept of a database machine to knock cost out of the layered technology stacks that most often address data-intensive applications. Epsilon tested NPS on its handling of a heavy load of simultaneous queries from Business Objects, Unica marketing management and other tools.

The company was satisfied with the speed and throughput; production query systems that took hours in batch with Epsilon's older systems could execute in minutes, says Coakley. "Now, our systems can answer multiple questions per day from our clients, which reduces the cycle time and ultimately how long it takes to get campaigns up and running." Coakley likes Netezza's price point: "We may be able to offer services that normally require large, expensive systems to organizations that don't want to spend a lot of money, such as regional banks and midsize firms." Competitive offerings from NCR/Teradata, Oracle and other major database providers will likely join Netezza and draw interest from organizations that currently view customer intelligence as beyond what they can afford.

CDI: CRM With a View

Along with query performance, the data warehousing and integration infrastructure is responsible for delivering that most coveted enterprise resource: the single view of the customer. Beyond multiple CRM and database marketing applications, most organizations have a combination of ERP, order management, call-center management, sales force automation and other systems, each of which owns important data about customers. External data service providers with demographic, credit and other information are also part of the puzzle.

Pulling relevant data together into a 360-degree view of a customer requires overcoming many integration obstacles: duplicate data, mismatching customer codes and attributes and so on. While some organizations pursue the "single view" to focus on products, financial metrics or regulatory compliance, improved customer intelligence is typically the main reason. That interest explains the current market buzz about customer data integration (CDI) and enterprise information integration (EII) as potential solutions to the need for a single view.

CDI covers a range of possible solutions, including centralized semantic stores, which put rules and validation programs to resolve integration problems into an application program or Web service available to all systems; master reference data stores, which consolidate customer information into one source of truth; and even the more well-known enterprise data warehouse (EDW) and operational data store (ODS) systems, which can segment, partition and aggregate customer data. Whereas the EDW traditionally serves the predictable information analysis needs of BI reporting, an ODS offers a subject-oriented integration of more volatile operational data. DWL, Initiate Systems and Siperian are prominent pure-play CDI vendors. IBM, Oracle, SAP and Siebel have CDI solutions in their portfolios. In January, Hyperion announced that it's acquiring Razza Solutions, maker of master data management (MDM) software. MDM is a technology flavor similar to CDI.

A key issue drawing attention to CDI and EII approaches is timeliness. Data warehouses are typically updated in batch, but real-time pressures are changing things. Streaming data is a new option; you can "stream" or "trickle" data into the EDW or ODS to continuously update single rows in the database, for example. Message-oriented middleware (including its Web services-oriented offshoot, enterprise service bus) delivers a stream of updates to "subscribing" data warehouses. With streaming, however, IT's challenge is to bring transaction-oriented management to the warehouse environment.

EII versions of CDI focus on delivering a single view of data, rather than the data itself. Avaki, Composite Software, Group 1 and MetaMatrix are leading pure-play vendors. With EII, you don't stage the data in an EDW or ODW; instead, the EII plays a middleware role to deliver integrated results sets to front-end BI or other user tools. Life Time Fitness, for example, uses Composite Information Server to join information from disparate data sources, including two large databases running with Microsoft SQL Server, another Java-based application and a hosted POS system. Life Time gives business users a consolidated view of each customer, without incurring as much data integration infrastructure complexity.

Harrah's Operational CRM

Harrah's Entertainment establishes its single view through an EDW running on Teradata. "We use the EDW to track customer information and activities so that we can interact with customers more intelligently in the future," says Sam Dillard, Harrah's IT director. "The EDW started out as a marketing idea, but we designed it to support multiple subject areas for better alignment of business units. Also, we can view ourselves in multiple ways. While we see ourselves as a consumer marketing company, we're also a retailer. Many of the performance measurements we take are what retailers use."

The company is renowned for its Total Rewards and Player Contact System loyalty programs. Harrah's provides selected customers with loyalty cards, which they swipe at gambling stations to gain Harrah's points. Meanwhile, Harrah's is collecting valuable data, which it uses to track customer spending and feed algorithms and rules systems that deliver guidance to employees (or self-service applications) about special offers they might provide. "Our analytic systems allow us to make smart decisions and help employees deal with our customers in a way that's rewarding for them. We have happy and more productive team members as well as more satisfied customers."

A key differentiator is an "active" approach to data warehousing, which moves Harrah's beyond analytic to what it calls "operational" CRM. "We take the analytics and tap into them in a real-time manner, instead of as part of a batch process," explains Dillard. "We pull historical data from the warehouse and use it with real-time data coming out of our operating environment." Harrah's uses Tibco business integration software adapters to provide a constant trickle feed of data into and out of the data warehouse. "To guide decision-making, our system looks at historical data for that customer or process and applies a business rule to make a recommendation of what the next step should be," Dillard says. "Our platform also lets our people do analysis on data much sooner than they otherwise would."

Future Foretold: Predictive CRM

With data on almost 29 million guests in its customer database, Harrah's has a prime resource for doing predictive modeling and analytics. "It's fairly easy for us to reach out to a segment in our database to see how a certain group has behaved and then predict how someone new would behave," Dillard explains. With help from Cognos software for predictive modeling, Dillard says Harrah's can "estimate what a customer's enterprise value would be to us."

Predictive analytics, or the application of data mining algorithms to discover patterns in data is already a competitive advantage for large organizations on the cutting edge (see the accompanying Field Report on Churchill Downs). IDC predicts that the software, offered by Computer Associates, KXEN, SAS, SPSS and others, will steadily attract customers in 2005. What makes predictive analytics different from traditional BI analysis is that you're letting the data do the talking. As opposed to BI queries that specify desired results, predictive analytics software leads a discovery process. The algorithms tap carefully managed data warehouses and other resources to uncover patterns and data relationships that help companies predict their most profitable customers, how pricing and product packaging might affect buying decisions and much more.

Of course, for customers, the future is right now. Predictive analytics, operational CRM and other methods are at their best when they enable companies to deliver a richer customer experience tailored to their immediate interests. With customer loyalty assured, the future will always be bright.