Matrix Meets 3rd-Generation Needs
Three years ago, it was the "next big thing." Elaborate systems would pump out hundreds of daily promotions, following carefully scripted, multistep contact strategies driven by customer behavior. Every company would need this software to compete. The market would be huge.
There are many possible reasons the future failed to arrive. Everyone was distracted by the Internet. Customer relationship management shifted the focus to inbound communications. Benefits never justified the cost. Organizations could not handle the change. It was too complicated for most marketing departments. But while the causes are murky, the effect is clear: The market for high-end campaign management software has remained quite small. At best, a few hundred such systems are running today, and it is unlikely that 100 more will be sold this year.
But though the market is small, it is still a market. And the few companies that do need such systems have requirements that only a few products truly meet.
These requirements relate primarily to scale. Setting up a 100-segment campaign takes an interface that lets you see the complete structure on one or two screens, then drill down to the details of each cell. Efficiently executing such a campaign takes specialized technology. Managing 100 campaigns a month takes administrative and automation features to complete them on schedule with a reasonable amount of labor. Developing, budgeting, approving and tracking that many campaigns takes project management and reporting features to coordinate dozens or hundreds of employees and vendors.
DataLodgic and N.Que (Matrix Technology Group, 727/669-7000, www.mxtg.net) are products for high-end campaign management. They are descended from Paradigm Ovation, a system originally built for an integrated communications agency. Given that heritage, it is no surprise that N.Que is a very sophisticated project management system while DataLodgic provides the traditional campaign management functions. The two systems share reporting tools and data access technology, with tighter integration planned over time.
Like most modern campaign management systems, DataLodgic separates creation of file segments from design of the campaign itself. The segmentation tool can access any customer database, either directly or through models that hide the physical structure. Users build a layered segmentation tree describing multiple file segments. Segments can be split on data values, user-defined calculations and random or ranked samples. The system can automatically create separate segments for each value of a variable, such as a state code, or build equal-size segments on continuous variables such as model scores. Both features can save considerable effort in complex segmentations. The user interface cannot build queries that require multiple SQL passes, such as correlated subqueries or comparisons to aggregate values, although users can insert their own SQL statements if they know how to write them.
Once a segmentation is built, DataLodgic can count the number of records in each segment and display a sample of the records themselves. This helps marketers validate that they have defined their segments correctly. The system then saves the segments in a library, where they can be used in campaigns. Segment definitions are usually saved as SQL statements, although the list of selected records can be saved as well.
DataLodgic campaigns are built as a sequence of tasks on a multilevel, collapsible tree. Tasks are created by a technical user and placed in a library accessible to others. A task may represent a specific communication such as a direct mail piece, a list extract, a data management process such as moving a file or a segmentation to send records to different branches in the campaign flow. When a task is set up, its builder defines the attributes such as cost and response estimates, output formats and offer details. The system comes with a set of common tasks pre-built.
Tasks also can include Visual Basic or SQL scripts to run external processes or send messages to N.Que. Model scores can be applied to a single record in real time or to groups of records in a batch process. Simple scoring formulas can be built directly within the system, or it can import SAS scoring scripts and execute these without calling SAS itself.
Execution of each task is controlled separately. The system provides a rich set of choices, letting tasks run on a regular schedule, at a specified interval after a prior task, after a database change or based on the result of a SQL query or script. The system keeps an internal list of records at each stage in a campaign, although this must be copied to an external database to maintain a permanent contact history. Reporting includes Gantt charts with start and stop dates for individual tasks and for campaigns as a whole, as well as campaign return-on-investment analysis for forecast and actual results.
N.Que tasks represent the steps needed to build and manage a campaign. They are separate from DataLodgic tasks, although each system can trigger a process in the other. N.Que provides a full set of project management functions, including task dependencies, critical path analysis, individual workload management, budgeting, cost capture, overdue task notification and automated rescheduling when plans change. The administrative portion of the system runs on a Windows PC, but workers can access their personal task list and enter billable hours through any Web browser. The system also provides extensive Gantt charts and tabular reporting on project status, costs and performance against plan.
DataLodgic and N.Que run on Windows NT/2000 servers and work with standard relational databases. The vendor also provides technology to read data in nonrelational systems. This is used largely to look up reference information in corporate operational systems, such as details of available offers. This can be a critical problem in large installations.
The products were released in 1999 and have about 10 clients between them. Pricing is based on the number of users and starts at $150,000 for five users for DataLodgic, and $65,000 for 10 administrative users and 50 workers for N.Que.