For Beta or for Worse

AS THE ROLE of customer relationship management in direct marketing increases, so does the number of CRM software and systems used to support all facets of it.

For this roundup of new or newly upgraded software systems, DIRECT looked at three firms: two that specialize in elements key to companies interested in keeping their customers satisfied-retention and fulfillment/ inventory control-as well as one focusing on a staple of database marketing, prospecting.

As in past roundups (DIRECT, July and December 1998), we asked beta testers and early adapters to offer reactions to and potential modifications for the packages reviewed.

ProspectSmart! ProspectSmart!, a new-customer forecasting package from Denver-based TargetSmart Inc., combines Experian customer data on two compact discs with the software’s mining, messaging and deduplication capabilities, as well as a platform that integrates with printing, fulfillment or call center operations.

ProspectSmart!’s Automated Profiling System allows prospect lists to be created from current customer files, and can be done either through a series of preprogrammed modules or customized characteristics. The profiling systems can create prospect lists using specified geographic areas, or by creating “grids” of high customer concentration, which are used to isolate targets in the same location.

The program also contains Creative Wizards that walk marketers through generating direct mail collateral and telemarketing scripts. The connectivity between the prospect data and the Wizards allows a high degree of personalization.

There are, however, limits to the amount of data that ProspectSmart! can provide while remaining easy to manipulate. The two discs do not include behavioral data, and given Experian’s purchase of Metromail (and therefore its BehaviorBank), there is a wealth of higher-level information that either needs to be overlaid or can’t be used in making marketing decisions.

But for mid-level marketers that target based on geography and basic household characteristics, the flexibility of the program is a more-than-fair tradeoff.

For CommNet Cellular Inc., a growing cellular telephony provider, prospect lists-especially for its two most significant markets-proved difficult to come by.

Initially, the company was working with list brokers, having only moderate success at trying to generate prospect lists. Often, the brokers were unable to provide names within the company’s stringent geographic requirements: It is currently licensed to provide service in Mountain and Western states, and has chosen to limit its marketing focus largely to rural areas.

“We had to teach them about our business, and give them copies of our customer database so they could dedupe [the prospect lists],” says Ginger Hauser, dialer administrator of the Englewood, CO-based company. “It was very painful. It could sometimes take weeks or months.”

It was also more expensive. After segmenting, formatting and loading charges, a name could end up costing between 20 cents and 30 cents. Using ProspectSmart! and buying names from the parent company on compact disc has reduced that to about 10 cents per name.

But cost was not the only factor that caused CommNet to bring the processing in-house.

The output files the company received were often difficult to transfer into its automatic predictive dialer system. By using ProspectSmart!, CommNet is able to define on a daily basis the type of targets it wants to solicit, based on factors such as ZIP codes, age groups, income levels or family makeup.

Once the names are loaded into the dialer, the company can then segment the lists between the 10 members of its in-house call center.

“We can say, you five are going to work on the older folks,” says cellular product manager Steve Erickson. “There is a higher percentage of people in the prepaid product category in the younger group, and we would talk about the approach [to be used when calling them].”

Inventory Forecast Management Like most retailers, catalogers face having to balance overstock-which can create the need for liquidation sales-against limiting on-hand supplies-which can result in back orders leading to cancelled sales. To avoid this, Smith-Gardner & Associates Inc., Delray Beach, FL has introduced the Inventory Forecast Management system, a module of MACS, its Mailorder and Cataloging System.

One template allows merchants to quickly view inventory requirement spreadsheets by item number, vendor, lead time necessary for restocking and other key information.

Another tracks weekly overstock quantities, incorporating historical inventory depletion and projected purchase order replenishment rates.

A budgeting mix-and-match feature allows inventory modeling for newly introduced items; depletion patterns for up to 10 related items can be combined and used as the basis for analysis.

A fourth feature allows a user to review purchase orders with a few keystrokes.

The primary function of the system is to generate a performance curve over the life of an offer. Response curve projections based on the sales and returns of similar products or styles, performance of other offers within the product’s category, or response patterns for similar offers are used to anticipate inventory needs after a given offer goes out to its audience. Response curve projections can be updated with actual numbers, causing the inventory requirements to be reconfigured. (Projections can also be stored separately, allowing the marketer to compare the performance curve of actual sales with predicted sales.)

For direct marketers that do one-off solicitations, the response curves are based on product performances in similar categories. Catalog retailers can use the function to anticipate need over a continuous series of solicitations, either on a seasonal or year-round basis.

At Catalog Ventures, Chelmsford, MA, Bob Piro, vice president of merchandising and operations, is testing the system. While he is currently buying enough merchandise to cover a season’s demand under the company’s extant inventory control method, he is using MACS Inventory Forecast Management to model a balance between back orders and overstocks, and plans to convert to IFM by fall.

MACS’ forecasting functions, Prio says, result from its predictive use models, based off nightly inventory updates.

The reports generated allow him to split his purchase orders, buying half of his inventory up front, with another half held for confirmation.

If sales are slow, Piro says, he then has the option of canceling the additional purchase. But if they are on schedule, he has the option to have the mid-season replacement orders delivered.

In the case of entirely new offerings, Catalog Ventures uses 18 months of purchase data from similar products-ranging from quarter-to-quarter sales figures for merchandise that is always offered to year-over-year sales for seasonal merchandise-in its analysis.

While Piro is careful to take advantage of bulk buying discounts, even with the split orders, the new system should result in savings to Catalog Venture’s warehouse supply space costs, and will allow the company to re-allocate capital that would otherwise have gone toward larger up-front purchases.

The new system will also allow Catalog Ventures to bundle orders for the company’s four catalog lines. Although each one features discrete merchandise, often vendors supply to more than one.

The warehousing of inventory information across all catalogs allows the company to make a single bulk purchase, rather than having to deal with four sets of order information and inbound freight costs.

Top Dog Decision Software Inc.’s campaign management and data mining tool is TopDog, a product whose core competency is its run time.

TopDog has the ability to make customer or prospect selections based on one pass through a database, regardless of the number of defined segments.

Continuing the canine theme, queries can be stored and reused through the software’s “stay” and “retrieve” functions (the “fetch” command having been co-opted by DataFetch, a function of TopDog’s “Golden Retriever” utility program that aggregates data and allows users to pick data field and sorting criteria.)

Version 6.0, the newest release from the Landover, MD, firm, has added features that make plumbing the depths of a data warehouse easier. The new campaign management capabilities include improved report sharing ability, enhanced mail file options, promotion history templates and campaign editing capabilities.

The Chicago headquarters of the American Medical Association uses TopDog to power its membership solicitations and book and product solicitations. Campaigns range from small, targeted lists to 600,000-name telemarketing or catalog solicitations.

For its product lines, the AMA uses regression techniques based on RFM variables to build a model. Once ranked, the organization’s telemarketing and direct mail efforts are targeted based on ranking provided by TopDog.

“One of the biggest [product] campaigns is with our reimbursement product,” says the AMA’s director of decision support systems Jonathan Van Oss. “It was a telemarketing campaign to 120,000 prospects for an insurance reimbursement procedural terminology product.”

The May campaign, says Van Oss, was “very effective,” but the software’s value came in the ability to act on recency. “If we didn’t have TopDog to do the processing,” he says, “we might still be able to do it but it would take a lot longer.”

But like many membership-based organizations, the AMA uses its segmentation modeling to beef up its retention efforts. The association is currently developing a predictive model of physicians most likely not to renew their memberships.

“They are younger, just out of residency, and in some cases more likely to be female than male,” says Van Oss. For them, the AMA plans to highlight group practice regulations and student loan repayment assistance programs.