Measuring Across Sales Channels
Many marketers still view each channel in a vacuum, judging its results only on attributed sales and promotion costs. Working with catalogers who have stores and e-commerce sites, we've learned important lessons about how these channels interact.
Retail's effect on mail order. Catalogs selling products that heavily overlap those offered in company-owned retail outlets see mail-order sales per thousand drop 20 percent to 45 percent in store trading areas. This falloff is about the same for previous mail-order buyers as it is for prospects on rented response lists. Mail-order falloff is quick, occurring within four to 12 weeks of store opening. Most importantly, mail-order sales rarely rebound.
Companies that collect and mail retail buyer names find them producing negligible mail-order sales. At best, direct sales barely equal catalog mailing costs. The exception is for the typically few such buyers who live outside a store trade area. Their mail-order sales can be four to 10 times those living near a store, high enough to be profitable on their own.
Not surprisingly, a positive correlation exists between distance to the nearest store and mail-order sales. The greater the distance to the closest store, the less a retail presence depresses mail order. Retail impact zones depend to an extent on the number of outlets in a market and on how much of a shopping destination they are. Mail-order sales typically experience substantial falloff as far away as 25 to 50 miles from the nearest store, farther than most managements imagine.
Mail order's effect on retail. Offsetting the harm that retail presence has on mail-order sales is quantitative and anecdotal evidence that mailings to retail and mail-order buyers drive incremental retail sales. Determining this amount, however, can be difficult.
Unlike mail order and e-commerce, which are driven heavily by promotional activity, some store sales, especially to prior retail buyers, occur regardless of such promotions or because of unrelated retail advertising. The challenge is to isolate the portion due solely to direct contacts.
Further complicating measurement is the need to attach a name and address to most retail transactions to match them to customers sent mail or e-mail promotions. Except for retailers whose customers pay mostly via a proprietary store charge, capturing this detail is a significant, costly and ongoing job. It is, however, doable.
Work done with firms that capture retail buyer names and addresses for most store sales shows the cost of mailing catalogs to some but not all retail buyer segments is justified. Justification is defined as profit on incremental retail plus mail-order sales exceeding mail cost. The best way to measure this is to track the retail purchases of like groups of past store buyers, only some of whom are mailed.
Analysis also shows retail buyers segment similar to mail-order names. More recent, more frequent, higher-value shoppers spend much more than infrequent, less recent names, with frequency a much better predictor of purchase than recency. Buyers who purchase from multiple channels are invariably the best customers, though they typically are only a fraction of a total file.
Use segmentation tools, whether RFM or more sophisticated models, to determine which retail buyers to contact and how often, just as with mail-order buyers. Experience suggests it is cost-effective to promote one-time retail buyers for far shorter periods than like mail-order buyers. It also appears that the positive retail sales effect of catalog mailings can be achieved with fewer contacts per name than are typically desirable to maximize mail-order sales and profit.
Retail sales of prior mail-order buyers living near stores form a much larger portion of the total than do the mail-order sales of retail buyers. While retail sales of mail-order buyers and prospects in trading areas often equal 50 percent to 75 percent of their mail-order level, mail-order sales of prior retail buyers total only 5 percent to 10 percent of retail sales.
E-commerce. Given the newness of e-commerce and that the Web serves as a distribution and an order-placement channel, similar cross-channel impact rules are only now emerging. However, mail/retail cross-channel experience argues persuasively that mailers determine the extent to which:
· Traditional mailings to mail-order and retail customers and to prospects drive recipient e-commerce sales.
· Buyers acquired via e-commerce respond to direct mail promotions and to a brick-and-mortar presence.
· Previous buyers from any other channel respond to Web promotions.
· Site visitors who are not prior customers are driven to place initial orders by phone.
Since buyers from e-commerce sites must give a name and address, tracking sales between mail and e-commerce channels is no more complicated than deduping e-commerce buyers against house and prospect lists mailed. While e-commerce buyers mailed catalogs should be given unique source codes, always match site sales against mailing lists as source code capture rates at most sites is low.
Any analysis of how different kinds of e-mail affect mail-order or retail buyers is incomplete without comparing total sales from all channels, of groups sent such mail, against a control group not sent the message. A simple way to track sales that sites drive to traditional order-placement channels is a phone number unique to the Web site.
For catalogers to make optimal decisions about allocating resources among available distribution channels, they must understand the extent to which promotions in each channel affect recipients' total sales. Failure to do so, and viewing each channel only on the basis of its attributed sales and costs, virtually guarantees lost profit.