This article originally appeared on the BeyeNETWORK
Bill Inmon’s recent five-part series, “Competitive Advantage and the Data Warehouse,” presented a compelling argument that, “[T]he data warehouse can be used in many ways to analyze the segmentation and growth of customer categories … The changes in categories and the reasons for the migration of customers between categories are at the heart of corporate strategy.”
Bill defined and described four categories: Profitable, Sustaining, Unprofitable and Entry. For example, net migrations from Profitable to Sustaining, and from Sustaining to Unprofitable, generally are cause for significant concern and should spur a forceful strategic response.
A logical extension of what Bill described is the construction of a CRM Growth Simulator from the atomic-level data within your warehouse, in order to identify, monitor and take advantage of the dynamics that drive the fortunes of your business. A CRM Growth Simulator works in tandem with the data warehouse to inform both strategy and tactics:
- The warehouse provides an important historical compendium of customer events such as the purchase of a product or service, as well as subsequent trends associated with these events such as an upgrade to the terms of a service. All of this is within the context of CRM efforts, product/service offerings and other data that captures the influencers of customer behavior.
- The CRM Growth Simulator estimates future events by extrapolating the past trends identified within the detailed data in the warehouse, under different scenarios of CRM efforts and expenditures.
- The business plans derived from simulations serve as a benchmark to judge future performance.
Hence, the warehouse functions as a window into previous reality, and the CRM Growth Simulator as a view to the likely future state of things.
A CRM Growth Simulator’s projections allow the calculation of total estimated ongoing profitability, with a metric known as Enterprise-Wide Long Term Value (“E-LTV”). E-LTV is the sum of estimated long-term value across the entire customer base; that is, the discounted sum of all future profit flows. It represents the net present value of the firm given the input to the CRM Growth Simulator of a set of parameters and accompanying customer/inquiry/prospect behavior models derived from the data warehouse.
An Iterative Tool to Inform Both Strategy and Tactics
A CRM Growth Simulator allows the investigation of E-LTV in an environment that transcends the limitations of past marketing decisions. However, there are four important ground rules for such work:
First, do not confuse cause with effect. For example, the state of being a multi-buyer is the effect of a customer’s loyalty rather than the cause of the loyalty. Hence, employing “give-away” tactics to encourage conversion from single to multi-buyer status is not likely to increase loyalty.
Second, realize that observed customer behaviors such as attrition are a mix of what you can truly influence vs. what is mere self-selection. Customers (as well as inquirers and prospects) vary in terms of their intrinsic quality. Unfortunately, a data warehouse presents only a partial view into this intrinsic quality. For example, consider the extreme case of a woman who has to replace her entire wardrobe because of a fire. Under normal circumstances, she is anything but a “clothes horse.” In other words, her intrinsic quality as a long-term customer is quite low. Unfortunately, the data warehouse will reflect only her purchase activity, and not the motivation behind it.
It is likely that this woman will lapse, and that it will be because her intrinsic quality is low rather than because of any deficiencies in the firm’s CRM efforts. Attempts to prevent her attrition, or reactivate her, are likely to be uneconomical.
Third, recognize that behavior subsequent to a marketing effort is not necessarily the result of that effort. What must be isolated and quantified is the portion of subsequent behavior that actually resulted from that marketing effort. This is known as a true incremental effect. For example, rigorous analysis might uncover that only 25 percent of the customer events that occur immediately after email blasts are the result of those blasts.
In most businesses there is always some degree of walk-in “traffic.” Within certain channels this can constitute the majority of activity. Such activity should be properly attributed to previous efforts to build brand awareness; which, in turn, are typically driven by mass marketing rather than targeted CRM.
Fourth, understand that longer-term, time-based measurements must be employed to quantify true incremental effects. For example, ongoing cycles of well-designed longitudinal treatment panels must be planned, managed and tracked. “One/off” tests, which are all-too-common among data-driven marketers, are often flawed. This is because their apparent findings are hopelessly contaminated by the interaction of prior and subsequent marketing contacts.
Working with a CRM Growth Simulator, significant increases in E-LTV can be generated by:
- Moving away from intuitive, rules-based, “historical-tallying” customer segmentation approaches. Examples include “Total Dollars of Activity within the Past 12 Months” and “Current Breadth of the Relationship.” In their place, “future-estimating” techniques such as statistics-based predictive models must be implemented.
- Developing long-term contact strategies in which content is tailored to each customer segment, and with a level of intensity that maximizes the return on investment.
- Running each contact strategy through the CRM Growth Simulator and selecting those that are the most promising—that is, those with the highest expected E-LTV—for live longitudinal testing.
- Establishing as the new marketing standard the contact strategy that, by dint of the longitudinal testing, proves itself to be most successful.
- Executing this cycle repeatedly to achieve continuous improvement, with the ultimate goal being CRM Optimization that: 1) maximizes current cash flow, and 2) systematically allocates cash flow across investments in customer cultivation, reactivation and acquisition.
As different contact strategies are developed and tested, it typically is discovered that increased marketing intensity among top-performing customer segments represents a major growth opportunity. Examples of increased marketing intensity include:
- Additional efforts by the field and/or phone sales force.
- “Priming” types of contacts such as phone calls, postcards or emails in advance of—or reminders subsequent to—field sales force visits or major mail campaigns.
- Special 1-800 numbers for the best customers.
- Loyalty and points-based programs.
- Additional targeted direct contacts such as letters, postcards or emails; as well as increased page counts in collateral material and the judicious use of more expensive postage and shipping options.
- Supplemental investment in the resolution of customer service problems.
The following is a high-level representation of a CRM Growth Simulator developed for a multi-channel marketing organization that emphasizes data-driven techniques:
The vast majority of the detail represented in the actual CRM Growth Simulator has not, by necessity, been shown. Nevertheless, at the “30,000 foot level,” the three key conceptual components are represented:
First, the Prospect Universe from which all new customers originate. This is further divided into: “Space,” which are magazine ads; “Rentals,” which are customers for whom the company has paid a fee for the privilege of making a one-time contact; “Internet,” which are customers who come in via the web; and “Walk Ins,” which are customers in the form of new retail traffic coming into stores, sales offices, branches, etc.
Second, Inquiry Cells of various levels of quality for individuals who have asked for additional information but not yet purchased. These cells can be defined by business rules such as Recency (the amount of time that has passed since an inquiry was made), or by sophisticated statistics-based predictive models that systematically interrogate multiple factors correlated with conversion-to-customer behavior.
Third, Customer Cells of various levels of quality.
The following are some of the inputs derived from the data warehouse going into the CRM Growth Simulator to capture the dynamics of this multi-channel marketing organization:
- The actual quantities for each cell as of Day 1 of the selected “starting” fiscal year.
- The parameters surrounding each promotion to prospects, inquirers and customers. This encompasses all relevant channels such as mail, email and—when applicable—outbound telemarketing and the field sales force. Specific inputs include: quantity contacted, cost of the promotion, percent responding, average event value (such as merchandise price or policy size) and percent of unfulfilled event (such as merchandise returned or policies canceled).
- The velocity with which individuals drop to less responsive segments as the time between customer events increases.
- Similarly, the pattern of reallocation to better segments that occurs as subsequent customer events occur.
The Agony of Construction …
A CRM Growth Simulator is ready for deployment once the corresponding atomic-level data that captures the dynamics of the firm have been input, and the result is the generation of accurate detailed financial results across several recent fiscal years. However, “the devil is in the details” because an inaccurate CRM Growth Simulator is worse than no Simulator at all.
Accuracy is a function of the proper capture and modeling of the details of customer, inquirer and/or prospect behavior. Specifically, this is the careful construction of segmentation models and transition matrices across the defined customer, inquirer and/or prospect segments and their association with financials such as product or service margins, marketing expenditures and the cost of capital.
Hence, before constructing a CRM Growth Simulator, the underlying data warehouse must be interrogated for areas of deficiency. The warehouse must contain complete, accurate and detailed customer, inquirer and/or prospect history, including:
- Unabridged, time-stamped, atomic-level transaction data.
- When applicable, post-transaction activity such as cancels, rebates, refunds, returns, exchanges, allowances and write-offs.
- All promotional contacts, including field sales, phone, mail and e-mail.
- De-duped individual-level data that has been properly linked to the household level.
- Likewise, for firms with business customers, individual-level data accurately that has been linked to the site level, and site-level data to the company level.
- A surrounding systems infrastructure that is able to easily recreate past-point-in-time customer views (“states”), including statistics-based predictive model scores and other methods of segmentation.
Creating an accurate CRM Growth Simulator is a substantial task, especially for complex businesses that deal with significant numbers of customer, inquirer and/or prospect segments and a large variety of promotions across multiple channels. For example, one well-known multi-billion dollar company spans the retail, catalog and e-commerce channels and generates more than 650 promotions a year within catalog and e-commerce alone.
Creating a CRM Growth Simulator is particularly challenging in today’s overlapping world of customer, inquirer and prospect “touch points.” The incremental effectiveness of each contact must be accurately estimated. It is imperative to quantify the exact amount ofcannibalization across contacts and channels. For example, a typical Retailer must navigate:
- Targeted mail and e-mail.
- Free Standing Inserts (“FSI’s”); that is, what falls out of the Sunday newspapers.
- Run of Press (“ROP”) ads appearing in the main body of newspapers and magazines.
- Television and radio spots.
- The powerful impressions that occur every time a prospect or customer passes by a store.
… and the Ecstasy of Implementation
The rewards of a properly constructed CRM Growth Simulator can be profound. A robust Simulator generates realistic customer, inquirer and prospect flows from segment to segment, in a process that assumes an ongoing dynamic of its own. When combined with corresponding revenues and costs, the result is monthly profit flows; the analysis of which can lead to penetrating strategic and tactical insights.
With a robust CRM Growth Simulator, what-if analyses can be processed to answer important strategic and tactical questions such as:
- Has E-LTV been increasing, decreasing or staying roughly the same in recent years?
- What would happen if you could lower your variable product or service cost by 1 percent?
- Would the short-term reduction in profitability from a more aggressive new customer acquisition strategy pay for itself with incremental long-run profits? Specifically, what if your acceptable acquisition cost were increased by 5 percent, 10 percent or even 15 percent?
- What can be done to turn things around in the upcoming year? And, from a longer-term perspective, will it adversely affect E-LTV?
- How long will a positive event “echo” in your business (i.e., increase E-LTV) or a negative development depress future results (i.e., decrease E-LTV)?
- What can your CEO reasonably promise to Wall Street analysts? And, what has to be done to back-up what has already been said (and which perhaps should not have been)?
- How does E-LTV compare with your firm’s market value as determined by the combined price of all shares currently outstanding?
- How does E-LTV compare with an offer on the table to purchase your company?
These are among the many ways in which a CRM Growth Simulator, as a dynamic extension of your data warehouse, allows you to take full advantage of the data to inform both your strategy and your tactics. With a robust CRM Growth Simulator, you can drive significant increases in revenue and profit, and a corresponding dramatic increase in E-LTV.