Multiple analytical tools

Feb 01, 1999 10:30 PM  By

Despite a vast selection of analytical tools available today, many catalogers don’t stray from the traditional customer segmentation of recency, frequency, and monetary value (RFM). In fact, according to Catalog Age’s 1998 Benchmark Report on Lists and Databases (July 1998 issue), more than one-half of catalogers still rely mainly on RFM cells to make promotional decisions.

But integrating multiple analytical tools into your database program doesn’t have to be as complicated or as costly as you may think. The key to state-of-the-art target marketing lies in implementing two datamining initiatives:

* determining whom to mail, via statistics-based predictive models that estimate future purchase activity, and

* determining what to promote, using four quantitative techniques-tree analysis, demographic profiling, focus groups, and survey research.

To show how you can use these tools to better target your audience and your offer, we’ll use the hypothetical example of a cataloger with annual revenue of $30 million that sells apparel, jewelry, and gifts geared mostly to women. Until recently, this cataloger used RFM segmentation, classifying customers based on the following:

* recency, or the number of months since the most recent paid order. Each customer was placed in one of four possible monthly categories: 0-6 months, 7-12 months, 13-24 months, or more than 25 months since the last catalog order.

* frequency, or the total number of historical orders. Customers were classified by 1, 2, or at least 3 previous paid orders.

* monetary value, or historical average order size. Our cataloger created three equal-size groups: “low,” “mid,” and “high.”

This segmentation resulted in 36 (4 x 3 x 3) RFM cells.

The cataloger spent an average of $1.25 per catalog mailed. As it was willing to mail to breakeven, the cataloger mailed to those cells that were expected to bring in an average of at least $1.25 per piece mailed. Traditionally, remails generated half of the response rate of a first mailing, so the cataloger dropped a remail to those cells that were expected to generate $2.50 or more on the initial drop. Roughly 10% of the customer base was deemed unprofitable-in other words, was expected to generate less than $1.25 per mailing.

But the 36-cell segmentation wasn’t very precise. Customers who purchased 10 times, for example, were categorized identically to those who had bought three times, when in reality they had been much more loyal and should have been designated as such. Another drawback was that RFM leaves out so many other key data, such as merchandise categories; satisfaction indicators, including returns, exchanges, and allowances; and overlay demographics, such as age, income, marital status, and the presence of children.

Clearly, an RFM strategy that took full advantage of our cataloger’s comprehensive customer database would require thousands-perhaps tens of thousands-of cells, some of them consisting of just a handful of names. Then too, such a strategy would be unwieldy to implement, with many opportunities for error.

The beauty of modeling But statistics-based predictive models do not have these limitations, since you can include all potentially predictive customer characteristics without the sample-size issues inherent in RFM-in other words, you do not need to create additional cells each time you add characteristics to the model. Predictive models consider all such variables simultaneously, rather than in segments, resulting in a straightforward ranking of customers by predicted future purchase volume. Predictive models are more stable than RFM cells, they’re easier to implement, and, they do a substantially better job of determining future purchase behavior.

The chart on this page shows the performance of a predictive model that replaced the midsize cataloger’s traditional RFM segmentation strategy. Remember that this cataloger promotes to $1.25 per piece mailed and experiences a 50% falloff with remails. All customers have been scored by the model, ranked from highest to lowest predicted performance, and grouped into 10 equal performance segments, or deciles.

Customers in decile 1 are expected to generate $8.14 per piece mailed, while those in decile 10 are predicted to bring in just $0.44. Combined, all the deciles average $2.49. The “lift” column shows the performance for each decile compared with this $2.49 average, while the “cumulative lift” column records the “running” lift of this decile and those above it. For example, deciles 1 through 5 have a cumulative lift of 158. This means if the cataloger limited a promotion to just this group, the results would be 58% better than the $2.49 average for the entire file.

The chart also illustrates the incremental power of the predictive model vs. the traditional RFM cells:

* Deciles 8 through 10 rank below our $1.25 breakeven, so they should be eliminated from future mailings. Our cataloger’s original RFM cells, on the other hand, identified only about 10% of the customer base as unprofitable, because RFM is a weaker segmentation tool and isn’t as effective in isolating extremely profitable or, as in this case, extremely unprofitable sectors. With the statistics-based predictive model, the money that is freed up by eliminating unprofitable circulation can be allocated to profitable catalog remails, as well as to innovative targeted marketing programs.

* Deciles 1 through 3 can be remailed. Even with a 50% performance drop-off, dollars per piece mailed will still exceed $1.25. RFM cells were able to identify only about 15% of the database as remailable.

* Decile 1 performs so well that it can be remailed twice. That’s because the resulting performance-even after a 75% drop-off-remains comfortably above the $1.25 breakeven. The RFM cells were unable to identify any customers who profitably could be remailed twice, again because RFM is less effective in isolating extremes.

The four quantitative tools There’s more to state-of-the-art database marketing than optimally identifying the mailable portion of customers. You also need to determine what products to promote, via customized catalogs, using additional analytical tools such as tree analysis.

Statistics-based predictive models, for all their power, create what is known as heterogeneous segments- that is, segments containing individuals with multiple characteristics. Decile 1 of our midsize cataloger’s predictive model, for example, contains male and female, younger and older customers who have bought many combinations of merchandise. In fact, these customers have just one guaranteed similarity: They are expected to order an impressive amount of merchandise in the future.

Tree analysis creates homogeneous segments; that is, each segment contains individuals with identical purchase or demographic characteristics. For instance, one tree analysis segment may consist of 40- to 50-year-old female jewelry buyers who had made four or more purchases averaging $500, with at least one of those purchases being made within the past six months. Another segment may consist of 30- to 35-year-old male electronics buyers who had made three or more purchases, at least one of them within the past 12 months. Marketing insight from the tree segments can help you tailor your catalog to the demographic characteristics and product interests of multiple customer groups. To identify multiple types of customers, our cataloger ran a tree analysis off database fields limited to merchandise categories and demographics. These fields are the most likely to offer clues on how to match the catalog offer to a customer’s needs. Other variables, including recency, frequency, and average order size, help predict purchase behavior but provide little insight into customer lifestyles or interests.

The results of our cataloger’s tree analysis are illustrated in the chart above. We see that female customers generate $2.88 in sales per catalog mailed, more than twice that of the $1.42 generated by male buyers. But the analysis also reveals that a small pocket of male customers who bought jewelry actually generate more revenue than the average female buyer.

The cataloger set out to uncover additional clues about the purchase dynamics of this profitable customer segment, starting with an analysis to determine the types of jewelry bought. In almost all cases, the merchandise bought was targeted for females. The cataloger developed two theories:

* These purchasers are women using their husband’s credit cards.

* These purchasers are men buying gifts for the significant women in their lives.

Our cataloger’s next investigative step was to send its customer database to an overlay company for demographic and psychographic enhancement. Scores of fields were then added to each customer’s record, including age, income, marital status, and presence of children.

The results suggested that these jewelry-buying households-regardless of whether the purchase is being made by a card-toting wife or a gift-giving husband-are leading “Ozzie and Harriet” lifestyles: They are families with children, living in single-family suburban homes, with professional, technical, and managerial occupations.

Knowing that these are married suburbanites rather than single city-dwellers would help in tailoring catalog copy and offers, but our cataloger still didn’t know if the purchaser is the husband or the wife. To gain a definitive answer, the company commissioned a series of focus groups as well as some comprehensive survey research. (Depending on your internal resources, such a research program might cost you about $10,000.)

The focus group and survey research determined that the majority of these individuals are “unimaginative male gift-givers,” men at a loss for what kinds of presents their wives might find appealing.

To fully leverage these findings, our cataloger formed a task force of marketing, creative, and analytical employees to develop a loyalty program to appeal to this group of customers. Based on the insight gained from the research, the cataloger built the program around the following key features:

* Automatic reminders for upcoming birthdays, anniversaries, and other personal gift-giving milestones. This is a registry based on self-reported information so that the cataloger can remind participants via phone or e-mail of upcoming special events that will require a gift.

* A consulting service to help participants with gift selection. The cataloger staffed its inbound call center with specially trained “gift consultants.” This service was tied to the existing database, to ensure that the c onsultants do not recommend duplicate gifts.

* All gifts are wrapped and then mailed with a card to the program participant or-if he’s out of town-directly to the gift recipient.

On the prospecting side, the cataloger’s circulation department began working with its list broker to identify male-oriented lists for mailing prospecting catalogs, which included a description of the loyalty program and a form for signing up. The results of the program? Significant increases in the cataloger’s revenue as well as profit. By employing tree analysis, demographic profiling, focus groups, and survey research, the cataloger was able to identify, cultivate, and grow a small but profitable group of male customers.

A competitive advantage A common misconception among catalogers and direct marketers is that only the largest companies can justify the costs of using predictive models. But that’s not necessarily true.

In a typical case study, a midsize cataloger that invests about $25,000 in a predictive model will see an incremental $18,000 in revenue and an additional $68,000 in profit over using traditional RFM cells-after just one promotion. When you consider the gains you might see after a full year’s mailing cycle, the profitability potential with predictive models is significant.

It’s becoming more critical for catalogers to develop insight into what offers to promote and who to target; it’s also increasingly important to cultivate customer relationships by establishing specialized, targeted marketing initiatives such as loyalty programs. Implementing multiple analytical tools and statistics-based predictive models can help you better compete in today’s competitive catalog marketplace.