FreeRFM Adds to Dependable Formula
Really good results do require some art in choosing group boundaries, smoothing out random variations and sometimes adjusting for different product lines. But even these sorts of refinements can be implemented without statistical training or specialized software.
Ironically, the power and simplicity of basic RFM may have inhibited development of software for more refined RFM techniques. Why pay for someone else's software when do-it-yourself RFM is so easy? The one widely available RFM package, RFM for Windows, from database marketing guru Arthur Hughes, illustrates the point with its low price of $495.
FreeRFM (Dynamic Businesses LLC, 866/672-1106, www.freerfm.com) sounds like another low-cost RFM system. But the name is misleading: It's not free (scored lists cost 3 cents per name), and it does more than traditional RFM. Confusion notwithstanding, that's a good thing.
FreeRFM begins by calculating several measures beyond standard RFM, still using the same transaction data. One value is "tenure," or total time since a customer's first purchase. This is combined with recency, frequency and monetary value to generate a RFMT score used to rank customers by their likelihood of making a future purchase. The ability to rank individual customers takes FreeRFM beyond the cell-based segmentation of conventional RFM toward the ranking scores of regression models.
FreeRFM also calculates a table of "latency" and "conversion" values, which are, respectively, the average number of days between purchases at each frequency level (i.e., between first and second purchase, second and third purchase, etc.) and percentage of customers who make each transition.
Latency helps create an intriguing value called "expectancy," which is the expected number of days until a customer's next purchase. This is calculated by subtracting the number of days since the customer's previous purchase from the appropriate latency figure. For example, if the average interval between the fifth and sixth purchase was 30 days, and a particular customer made his fifth purchase 10 days ago, the expectancy would be 20 days. Expectancy is obviously merely a rough guess, but it does identify customers who appear ready for their next purchase or are overdue.
FreeRFM combines expectancy with frequency to assign customers to four life stages: new customers (single purchase with positive expectancy), loyal customers (multi-buyers with positive expectancy), potential defectors (multibuyers with negative expectancy) and lost customers (single purchase with negative expectancy).
Reports generated by the system show the number of customers in each group and average values by group for recency, frequency, monetary value, tenure, expectancy, average order value and lifetime value (defined as all past purchases less direct and indirect costs - direct costs are calculated as a percentage of each customer's past revenue while indirect costs are shared equally). Users can track changes in these values over time to understand how their business is doing. Online documentation explains each measure in detail and suggests appropriate marketing tactics for each life stage.
FreeRFM also uses customer transactions to calculate key performance indicators. These include the number of customers, orders, gross sales, profit per customer and average order value. The simplest report compares KPIs for the current versus previous year. A more detailed version breaks out current-year figures for new vs. previous customers and prior-year figures for retained vs. lost customers (that is, customers who bought last year but not since). A final report shows the year-over-year change in KPIs from retained customers and KPIs for acquired and lost customers combined.
Other reports project sales for the coming six weeks. The system estimates future purchases for each customer using purchase history, latency and conversion rates. Customers then are ranked using the RFMT score and divided into two, four, five, 10, 20 or 100 equal-sized groups. For each group, the system reports current recency, frequency, monetary value, tenure and lifetime value, plus projections for orders, gross sales, average order value, profit and profit per customer.
The profit figures are based on user-provided assumptions for costs of fulfillment, returns, service, marketing, product and overhead. Different types of costs are calculated per order, per customer, as percent of gross and as percent of net sales. Another report in the same format shows actual results for the prior six weeks.
The combination of business analysis, sales projections and RFMT ranking make FreeRFM more than a simple RFM system. It is aimed at small businesses with minimal inhouse direct marketing expertise. Its methods are most appropriate when customers make regular repeat purchases and behave similarly across product lines.
FreeRFM is very easy to use. It is hosted by the vendor and accessed through a Web browser, so no software installation is necessary. Users upload order files in a tab-delimited text format. The only required fields are order ID, order date, order amount and customer ID. Users may also load customer name, mailing address, e-mail address and phone number. Once the data are loaded, FreeRFM consolidates orders by customer ID, calculates latency and conversion, and generates RFMT rank, life stage, recommended marketing strategy, recency, frequency, monetary value, tenure, expectancy, average order value, lifetime value, and average order value for each customer. Users can load their data, view the system reports and see individual customer records for free. Output files, including all the values just listed, cost 3 cents per name. These are provided in comma-delimited text format.
FreeRFM is based on software originally developed for a direct marketer of church communication materials. An early version was launched in January, and the current version was released this month.