Back in 1985, a start-up firm called Marketing Profiles Inc. offered bankers the first desktop database marketing system. Max$ell used the technology of its day — the DOS operating system, character-based monochrome screens and a slow, proprietary
database with a fixed record structure — but it offered unprecedented data access, ease
of use and user support to marketing and research departments.
Although Max$ell looked increasingly dated by the early 1990s, sales accelerated
after MPI was purchased in 1994 by the John Harland Co. Harland, the country's second-largest check printer, marketed Max$ell through its sales force, pushing it to 1,100 installations.
Max$ell sales were helped by the virtual abandonment of the small bank market by its traditional competitors — Harte-Hanks, Customer Insight and OKRA Marketing — who developed new technology targeted to larger clients. This changed in 1996, when Harte-Hanks and Customer Insight both launched products aimed at small institutions. (OKRA was purchased by Harland in 1996 and continued to focus on larger clients.)
The country's largest check printer, the Deluxe Corp., also began selling a marketing database system in 1996. At the same time, the banks themselves were asking
for new capabilities to support closer integration of their marketing databases with branch and call-center activities. In this environment, Harland began work on a successor to Max$ell. It released the result in May.
Relationship Manager (Harland, 407/245-7600, www.harland.com) is designed to combine sophisticated technology with simplicity of use. Users can view data as it appeared at different points in time, at different levels of summarization and with different relationships among its elements — without being aware of how it is actually stored. The software also lets authorized users add new data elements, control data loading procedures, define consolidation and verification rules, perform calculations and create extract files — again without manipulating the underlying database directly.
The system does this by storing the date of every database change, which allows it to determine the value of any element at any time in the past. Queries and reports use this to show how information has changed over successive time periods and to compare a current vs. previous state for tasks such as response analysis or modeling.
Although Relationship Manager data is stored in relational database tables and is
accessed through standard Structured Query Language (SQL) statements, converting the
result into the specified timeframe requires a special data structure and non-SQL processes. Similarly, queries are generated in a specialized “set representation” and only later converted into actual SQL. These conversions and most other processing occur on the central server computer, so traffic between the server and desktop client machines is
minimized. Relationship Manager is intended to work with any database having a Microsoft Open Database Connectivity (ODBC) driver, although the initial implementation is limited to Informix.
Some elements of the data model must remain constant for Relationship Manager to perform its manipulations, but the system allows users to add data elements, set up multilevel hierarchies of products and organizational units and store calculations either in physical fields or as “virtual” fields whose value is generated only when they are used in a selection or report. It also can store individual transactions and contacts, track external events such as competitive offers and economic conditions and define multiple household relationships. Profitability calculations can use a different cost of capital for accounts of the same type — say, based on the age and term of a mortgage — addressing one of the primary criticisms of earlier marketing systems.
While setting up a selection or report, the system automatically presents data at the level of aggregation initially specified by the user. Similarly, when users are building a
selection, Relationship Manager only offers the operators and data elements that are
appropriate to the type of comparison the user has requested. This means users must specify the comparison type and operator before selecting the actual data element, a sequence that some may find difficult. Users can extract random or Nth samples from a
group and can limit selections to a specified quantity per bank branch or other entity.
This is very impressive technology, and making it simple for the user is still more
impressive. Ironically, the promotion and analytical functions that use this technology are less advanced. Selections are limited to a single file segment at a time, and the system lacks integrated graphics, scoring and mapping. While it offers a full set of marketing reports and user-defined report writer, the cross-tab is limited to two dimensions and doesn't let users create calculations on cell contents.
The system does a much more impressive job of sharing the promotions and reports once they are created. Information is stored in project folders, which can hold selections, reports, memos, graphic images, videos and contact definitions. These folders
can be assigned to a particular user, a promotion or a bank branch. This allows Relationship Manager to become a primary means of distributing information throughout the organization.
Reports can be saved as spreadsheet or .pdf viewer files, allowing them to be shared with people who aren't Relationship Manager users. Information about a marketing “contact” can include its contents, channel, output medium and the logic that will be used to identify responses in promotion analysis reports. Users can send selections and other tasks to a scheduler that will execute them at regular intervals or when specified conditions occur.
Eventually, users will be able to build and store their Relationship Manager databases in-house. But the initial release requires that databases be maintained at
Harland and accessed via telephone hookup. This has become a fairly common approach in the small bank market, where many institutions do not want to invest in software,
hardware or staff to run a system internally.
Pricing is based on the number of accounts. It includes a one-time installation fee
between $6,820 and $8,250 and a charge of $1,000 to $10,000 a month for monthly or
weekly updates. Harland is initially targeting institutions under $5 billion in assets.
David M. Raab is a consultant and author of “The Guide to Database Marketing Systems.” He specializes in the evaluation and selection of marketing technologies.