2005 Jul 01
Comprehensive Customer Metrics
David M. Raab
DM Review
July, 2005

As I said to a friend the other day, you know you’re in trouble when you start quoting yourself. But let’s face it, certain things do repeat themselves over time. In the area of marketing systems, one recurring topic is the need to help users make better decisions.

More specifically, the issue is that new systems often provide marketers with more options than they can effectively manage. A new channel like the Web–back when it was new–opened a nearly infinite range of possibilities for customer treatments. It took years for companies to get some idea of what makes sense. They had to invent new metrics, find new benchmarks, develop standard practices, understand and guide customer expectations, and experiment to refine their results. Sometimes they had to reverse direction: pop-up ads seemed brilliant for a time but then became simply annoying. Today, wise marketers use pop-ups rarely if at all.

Even in established media such as direct mail, new tools like advanced segmentation systems opened new opportunities for precise targeting and personalization. Marketers again had to adjust their practices to take advantage of these. Today they face similar challenges in optimizing real-time interactions across multiple touchpoints, a relatively new capability at many organizations.

So how do companies manage when faced with new situations? Of course, the true pioneers have no choice but to explore things by themselves. Most of them like it that way–that’s why they’re pioneers. But those who follow often want some help. This is why a strong services group that can help clients make use of new tools has been essential to the success of many pioneering marketing software vendors.

What’s different today is that the newest systems give companies an opportunity to manage not specific channels, but the entire set of interactions with each customer. This poses two related challenges: figuring out which policies really will be most effective, and deciding how to allocate resources across all available activities.

At a technology level, what’s happened is that communication and integration mechanisms such as XML and Web services have made it possible to view customer interactions within touchpoint systems, even if those systems were not originally open to external access. This means companies that didn’t invest in a comprehensive customer relationship management or enterprise management system–or who bought one but never successfully deployed it–can centrally capture interactions as if such a system were in place. Another critical enabling technology is the customer data integration hub. This correlates scattered information to recognize the same customer across different systems even without a shared identifier.

Identifying the most effective policies requires very detailed information. Visibility across different systems lets marketers assess the long-range results of any treatment policy: for example, how changing the mix of incoming customers affects retention rates and service costs in later years. Measuring the impact of everything on everything is not possible, but powerful analytical systems can identify important correlations. The key is assembling the detailed customer data in a central location for analysis. Careful test design helps where kitchen sink analysis–just throwing in all the data and seeing what relationships are present–is ineffective.

Allocating resources works at a much higher level. The goal here is not to identify the exact best treatment in each situation, but to set some general investment priorities. This requires assessing the average and, more important, the incremental value generated by expenses in each category. It implies some way to measure long-term customer value–typically discounted cash flow on future profits, but preferably also including more subtle factors such as value of referrals. It also requires understanding the connections between resources: how spending more on customer service would impact attrition rates, for example. Using such data to identify optimal allocations requires simulating business results over a period of years.

Managing all interactions requires consistent metrics across different channels. These metrics include revenues and costs, non-financial quantities such as interaction counts, and standard classifications for the interactions themselves. Developing such metrics is not trivial: to take a familiar example, consider how long it took Web marketers to settle on unique visitors as a meaningful metric and the effort still required to extract this information from raw Web logs. Once the relatively few key metrics are defined, marketers face another, even larger task of mapping channel-specific information into the common categories.

Painful as it may be, such effort is required to answer even simple questions. Let’s say you want to know whether it’s better to offer renewals by mail or telephone. A proper answer requires aggregating the different types of costs from the two channels, gathering renewal revenues (probably from a different system than the costs), and tracking the later behavior–of all sorts in all channels–of customers treated one way or the other. Without gathering all these kinds of information, there is no way to really determine which treatment is more profitable.

Customer metrics are just beginning to emerge as a specialty distinct from general business performance measurement, operational metrics, and traditional, campaign-oriented marketing metrics. It will take some time for standards to evolve and even longer to build connectors that convert real-world data into the standard categories. But marketers who want to accurately measure the results of their customer management policies will find these metrics essential.

Of course, measurement is not an end in itself. The real goal is to use the information to identify optimal policies and refine them over time. Consistent measurements of interaction values can highlight misallocated resources and opportunities for improvement. A truly comprehensive set of measurements, covering all interactions throughout the customer life cycle, would enable marketers to identify all of their customer treatment options and compare the long-term results of each choice. Even though complete omniscience may never be achieved, the new customer metrics bring marketers much closer than ever before. Beyond omniscience lies omnipotence. The mappings used to aggregate operational data into metrics can be traced backwards to transmit improved policies from the central customer management dashboard to the interaction systems themselves. This is an even grander vision than comprehensive customer metrics. It would require yet another layer of technology to accomplish. But it stands as a worthy goal, since it would finally put marketers in true control of the full customer experience.

* * *

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics. He can be reached at draab@raabassociates.com.

Leave a Reply

You must be logged in to post a comment.