How Data-mining Can Help Drive Personalized E-mail

Personalization and one-to-one marketing have become hot customer relationship management (CRM) topics in recent years. Indeed, glowing case studies from the direct marketing industry abound. Yet the core concept of personalizing print promotions has been slow to gain acceptance by catalog marketers in particular. Why?

Their reluctance may be due to past experiences in the traditional print catalog world, where the incremental cost to vary the offer can be prohibitive. As a result, most print catalog promotions deliver a product-centric message. To be sure, that can work for some catalogers, but it still doesn’t tap into the notion of true one-to-one marketing.

E-mail, however, provides a good vehicle for delivering — at relatively low cost — targeted, customer-centric, promotional messages to subscribers and opt-in customers. Previously, versioned messaging entailed simply plugging each recipient’s name into the body copy of an e-mail. But through the use of today’s automated e-mail-deployment systems, your messages can be varied to reflect the specific interests of your various recipients.

For example, Best Buy recently segmented online buyers into two groups based on their past purchases. Those who had bought stereos and CDs were segmented into the music group, and those who had purchased DVDs, TVs and VCRs went into the movie group. Best Buy then sent music-centric e-mail offers to the music group, and movie-centric offers to the movie segment.

Other ideas for personalized e-mail promotions: Before mailing a catalog, send an e-mail to alert customers that their new editions will arrive soon. Or send e-mails after your mailing to steer customers’ attention to specific offers you think they might like.

Through integrated campaigns of this type, your tried-and-true, product-driven catalog offers would be propelled by e-mail messages that identify with customers on an individual level.

By now you may be asking: “Aren’t there some privacy concerns with this approach?” Indeed, there are some. But more and more customers are accepting e-mail as a viable and reliable communications vehicle from trusted brands, so your properly executed and personalized e-mail campaigns should not raise an inordinate amount of recipient concern. After all, e-mail already is being used by many merchants for one-to-one customer communications about everyday transactions such as order acknowledgements and shipment notifications. The scope of the opportunity to personalize your promotional efforts should continue to grow as the number of new and existing customers who give you their e-mail addresses increases in time.

Mining for Relevance

The ultimate goal for one-to-one marketing is to develop and nurture customer loyalty by adding relevance to your e-mail communications, that is, by presenting customers with the right offers, at the right time and for the right reasons. To do that, you must uncover the logical customer segments for personalization. This means that some data mining into your housefile is in order.

The degree of success you can reasonably expect to achieve through the data-mining process is dependent upon the following:

• The scope and quality of the available data. To gain a complete view of your customers, you’ll need a central repository of all appropriate customer information across all channels. At a minimum, this information should include customer level summary data (e.g., RFM) and purchase-transaction history, including product-level detail.

To gain a more complete picture, append external customer characteristics (e.g., age, income level, household members, dwelling type). Other unique customer information elements such as promotional history and retail store proximity also should be included.

• Ease of access to the data for analysis. Once the proper information is in place, you’ll need a system that allows for access to your data so you can ask questions — and get answers — about your customers.

During the past few years, a new generation of high-performance, marketing database-access solutions (e.g., Alterian, SAS, Unica) have become available. These new systems give marketers fast, user friendly access to their customer data directly from their desktop PCs, at price points that are affordable and cost-justifiable for most mid- or large-sized catalog businesses.

• The right people involved in the process. Beyond having the right customer information and the appropriate software, you must empower the right people to mine the data.

Pam Maxwell, vice president of marketing at catalog company Interline Brands, put the data-mining process in proper perspective when interviewed for the November 2002 issue of Catalog Success. She said, “You need someone who knows how to mess around with the customer data to find interesting patterns.”

She advises fellow catalogers to hire people who can dig without knowing ahead of time what they’ll find. Oftentimes, analysts who are masters at the science of performance tracking and measurement are not practiced in the art of data mining. So you may need to add some input from the creative side to point out the right places to dig for possible patterns.

Where to Dig

When looking for opportunities to personalize, investigate the following areas within your customer information base:

• Prior behavior patterns. This involves looking at customer purchase transaction history to find groups of products that tend to be purchased by one segment of buyers. It may mean looking at products and product departments in a broader sense (e.g., all backyard items as a single category).

• Demographic, firmagraphic and geographic patterns. Segments may be defined based on life stages, socio-economics, climate, culture or other characteristics beyond purchase behavior.

• Motivation to buy. This also can be a powerful area of differentiation for personalization. For example, consider the difference in motivation to buy PCs within different types of households (e.g., on-campus college students, young professionals, families with children, empty-nesters). Most databases don’t capture customers’ reasons for purchase, so the motivation to buy must be inferred once the segments are defined.

Measure Performance

Several catalogers have told me they had tried e-mail personalization once on a prior campaign, and it didn’t work. My question to them is: “How do you know it can’t work if you tried it only once?”

The long-term retention of customers is what relationship marketing is all about. Given this objective, return on investment isn’t necessarily the proper measure of success. While an immediate increase in open rates, clickthroughs, order conversion and/or catalog response rates at the campaign level would be desirable and might be doable, the true success of a CRM effort can be evaluated only in time.

Personalization is only the tip of the iceberg. There are limitless opportunities to gain customer insight through the data-mining process. With the appropriate resources in place, you can:
- refine your customer list segmentation;
- analyze campaign performance;
- measure customer lifetime value;
- track retention/defection rates;
- manage customer contacts; and
- conduct product-affinity and market-basket analyses.

It all adds up to making effective use of your data to derive business intelligence that will impact your bottom line.

Keith Pietsch is vice president of analytics for Donnelley Marketing Catalog Vision, an infoUSA company. He can be reached at (952) 541-6548, or e-mail: keith.pietsch@donnelley.infousa.com.