Perish the paradox: using analytics to unearth true intelligence in customer interactions. (Call Center/CRM Management Scope).

Every year, XYZ Pharmaceuticals receives about 5,000 telephone orders from pharmacies and patient care facilities for medications to treat hypertension. Based on company experience over the past 18 months, XYZ knows that fully one-third of those callers will also inquire about nutritional supplements and home cholesterol tests. On the basis of this intelligence, XYZ's customer service reps discuss the company's nutritional and home testing products whenever they rake an order for antihypertensives. This "marketing campaign" costs XYZ nothing. But, by cross-selling to a customer segment that has historically demonstrated its interest, this call center-based marketing campaign boosts revenue for XYZ nutritionals and home testing products by an overall 25 percent.

Fact or fairy tale? The ability to learn from experience - to develop a store of knowledge that positions your company to continually optimize performance - is unquestionably the key to staying in business. Deriving maximum value from every bit of transactional data increases your return on investment in people and processes. It equips your business to better thrive in a competitive, crowded, and cost-conscious marketplace.

For companies whose customer interactions occur primarily through the contact center, the task is truly formidable. Their transactional data are square pegs for round holes; they are nor the structured content for which traditional databases have been built. It is in the air, over the telephone lines, on the Internet. It is in the tone of a customer's voice, an objection raised, a question casually asked. As your customers and suppliers continue to adopt new channels for communication, the volume of data and the complexity of capturing and analyzing those data also increases.

How, then, do we harness these contact center data to target opportunity across your enterprise and within your customer base? With automated enterprise analytics that capture and mine your customer interactions for accurate, meaningful intelligence and then deliver this intelligence directly to the people who need it.

So how intelligent is your recording system? What solution is better positioned to capture your contact center interactions than your contact center recording system? You already record for quality monitoring and risk management; why not heighten the return on investment by using your recording system to gather market-building intelligence about your customers and your performance?

When you consider the analytic component of your contact center recording system, should you be thinking of information - facts and statistics - or should you be seeking out intelligence? Facts are nice to note, but relationships that link cause to effect are more likely to enhance performance and profitability.

The Facts And just The Facts

In its simplest form, analytics take the shape of reports, charts and graphs. While illuminating, statistical data are often misleading and can point to a less-than-lucrative course of action, Let's look at how XYZ Pharmaceuticals fares using these traditional but limited analytic tools.

* The latest XYZ Pharmaceuticals quarterly sales report shows that sales of anti-hypertensives, nutritional supplements and home testing products continue at a slow pace, prompting the conclusion that these are non-growth markets for XYZ. However, the sales report also charts increasing demand for XYZ's eye care products, primarily from orders placed with XYZ reps visiting customer sites. Sensing a growth market, XYZ launches a direct mail marketing campaign to boost eye care product sales even more and instructs its telephone customer service reps to mention XYZ's eye care product line on every call, even if the order is unrelated to eye care products. Sales for XYZ's eye care products increase a modest 5 percent over the next four quarters, offset by the cost of the direct mail campaign.

* Relying on statistics that show a surge in demand for eye care products and stagnant growth in other product lines, XYZ Pharmaceuticals embarks on a costly marketing campaign that leverages a minor opportunity. Paradoxically, XYZ misses a more lucrative opportunity because its customer interaction recordings are nor considered in the analysis. This scenario illustrates how taking findings out of context may actually reduce the value of corporate data.

Clairvoyance And Lots Of Cash

Online analytical processing, a/k/a OLAP, represents a giant step in the evolution of analytics. OLAP yields meaningful information by assessing actual events under variable conditions, enabling businesses to determine if certain relationships and patterns exist.

Quantitative marketing scientists store corporate data in multidimensional OLAP cubes. In order to build an OLAP cube, these trained analysts must know in advance the sorts of relationships (and data) that will be considered: products sold by month, products sold by contact center agent, agent performance by region, agent performance by time of day, agent performance by product and so on.

Unlike reports and other static representations, OLAP puts the facts into a business context. The process, however, is complex and expensive and may still present paradoxical findings if the person asking the questions cannot foresee the right questions to ask.

* XYZ Pharmaceuticals has experienced steady demand for anti-hypertensive medications from a loyal customer base made up of pharmacies and patient care facilities. Seeking to boost revenue, the director of business development looks for other products that might appeal to these loyal customers. He knows that many patients with hypertension also suffer from high cholesterol. So, he asks XYZ's team of quantitative marketing scientists to determine if a relationship exists between orders for anti-hypertensives and home testing products. Two days later, the director of business development receives a report confirming this relationship, and XYZ's call center agents are instructed to discuss the company's home testing products whenever they take an order for anti-hypertensives. Over the next four quarters, revenue from XYZ's home testing products increases by 13 percent (less the time and cost of having trained analysts retrieve and analyze 18 months of call center recordings to confirm the director's hypothesis) . Sales of XYZ's anti-hypertensive and nutritional products remain stagnant, and the company even considers dropping its nutritional line.

* Since no one at XYZ had thought about the relationship between orders for anti-hypertensives and interest in nutritional products, the company's data analysts do not look for supporting data. Paradoxically, the company considers eliminating a potentially profitable product line and misses what could be a lucrative marketing opportunity. Regardless of the

outcome, the company spends significant time waiting for analyses and substantial budget dollars supporting a team of quantitative marketing scientists.

Mining For Gold

In a competitive environment where market share can vanish in the click of a mouse, analytical tools must offer optimum value at an optimal cost and with optimal speed. Reports, graphs and charts provide statistical information our of context, leaving users to guess about the real areas of opportunity and improvement. OLAP provides the business context and confirms meaningful trends, but it is expensive and time-consuming to engage; even worse, results are limited by the preconceptions of statistical marketing scientists and the people who request the analyses.

Data mining, however, offers businesses the capability to unearth the truth accurately, automatically and affordably. Data mining evaluates large volumes of data to unearth subtle patterns and cause/effect relationships within the business context and without prior prejudices regarding trends, relationships and future market conditions.

Data mining transforms unstructured, apparently unrelated data into intelligence that businesses can use to more fully leverage current business conditions and improve business performance. For data mining to be truly successful, however, it must produce intelligence that can be readily understood, used and shared -- intelligence that is actionable.

Every year, XYZ Pharmaceuticals receives about 5,000 telephone orders from pharmacies and patient care facilities for medications to treat hypertension. The analytics component of XYZ's contact center recording system finds that over the past 18 months, fully one-third of all callers who ordered anti-hypertensives also inquired about nutritional supplements and home cholesterol tests. The system automatically sends this finding via e-mail to the vice president of business development, who is not surprised by the relationship between orders for anti-hypertensives and interest in home testing equipment, but never suspected a related interest in nutritional products. XYZ's customer service reps are instructed to discuss the company's nutritional and home testing products whenever they take an order for anti-hypertensives. This no-cost "marketing campaign" -- based on empirical data automatically gathered, stored, analyzed and delivered to the vice president of business development -- boosts revenue for the comp any's nutritional and home testing products by an overall 25 percent cross-selling to a targeted customer segment.

XYZ's experience yields several important lessons about effective data analysis. Your corporate data should be comprehensive and accurate. Otherwise, they may send your marketing and performance initiatives in the wrong direction. Your contact center recordings are a truly valuable source of business intelligence; they tell you exactly what the marketplace is saying about your products and processes. But, while selective call recording may provide enough content for monitoring quality, only full-time recording can provide the richest, most complete and true-to-life sample of data on which to base your mission-critical strategies.

To more fully recreate the context of each contact center interaction, your recording system should incorporate information from the switch, from company data systems and from company customer relationship management databases. Only a recording system built on open, non-proprietary standards can provide this breadth of integration in essentially any hardware environment.

Your data should be accessible and affordable, and your data mining application should be up to the task. If you store your contact center recordings online in industry-standard file formats, your staff and analytics technologies can access these data whenever, wherever and for as long as they are needed. Plus, a contact center recording system that allows you to retain recordings with your existing storage infrastructure carries a lower total cost of ownership and operation.

Do make sure, however, that your data mining application can handle the truly massive volume of information captured by contact center recording. As new communications vehicles emerge and the role of the contact center continues to grow, your corporate data will also grow, and your data mining solution must scale to the demand.

Your data mining application should unearth the "truth" in your contact center recordings and maximize the value of the data you have collected. As the fictional XYZ Pharmaceuticals discovered, your marketplace may not always act or think in a predictable way. Your data mining application should give you answers to questions that you did not know to ask. It should help you identify your best customers, your best practices and your best opportunities. It should help you target areas for improvement across your organization -- from your agents to your supply chain to your processes -- so that you can optimize performance and profitability enterprisewide.

Your data mining application should deliver intelligence to the people who need it, in a form that they can readily understand, use and share. Intelligence is of little value if it cannot be applied to real-life situations or if decision makers do not know it is there. Your data mining application should provide intelligence in language and a visual format that the business user can easily understand -- without the need for trained analysts. Your data mining application should automatically "push" intelligence to business decision makers according to fulfillment rules that you define, so that the content is pertinent to the business role of the recipient and intelligence is automatically at hand as soon as it is available. Finally, your data mining application should present intelligence in a format that can be easily shared with other people and departments in your company, empowering your organization to make more informed, effective and profitable decisions and to improve performance enterprisewide.

Your contact center interactions keenly reflect your people and products in the eyes of your marketplace. Your contact center recording system should help you capture this vast store of data, enrich it with information from other enterprise management systems, and mine it for key insights that target areas of opportunity and growth. It is this "actionable intelligence" captured from contact center interactions and delivered to the right people at the right time across your organization that positions your company to excel in even the most competitive and contentious business arena.

Elan Moriah is president of Verint's Contact Center Business Intelligence Solutions Division (www.verintsystems.com). Moriah is responsible for leading the strategic initiatives of the division, including sales and marketing, engineering and research and development. Verint's Intelligent Recording solutions provide insight about customers, agents and processes, enabling decision makers to access this critical information at their desktops and use it to deliver a branded customer experience.