2005 Oct 01
CampaignRunner Inc. CampaignRunner
David M. Raab
DM News
October, 2005

It’s easy to doubt the future of campaign management software. Epiphany, Siebel and Doubleclick’s former SmartPath and Protagona have all been acquired in the past few months at unimpressive prices. SAS and Teradata sell their products as parts of an enterprise-wide offering. Even the major remaining independents, Unica and Aprimo, position themselves as comprehensive marketing automation solutions rather than plain old campaign management.

But like pesky mammals evolving in the shadows of dinosaurs, small-scale campaign management vendors survive and even flourish. Long-established specialists including Decision Software Inc., Alterian and smartFocus offer sophisticated campaign management at reasonable prices. They are attractive options for marketers who don’t want, or can’t afford, a broader product.

CampaignRunner (CampaignRunner Inc., 732-429-9458, www.campaign-runner.com) offers its own version of a dedicated campaign management system. Like all campaign managers, it allows complex segmentations of a customer file and stores selections in a promotion history. CampaignRunner does this using a single flow chart with nodes for data extraction, transformation, combination, segmentation, analysis and outputs. Some other campaign managers offer similar capabilities, but they usually break the tasks into separate processes. For example, it’s common to build segments independently of the campaigns that use them. CampaignRunner’s unified approach makes it easier to understand the flow of a given campaign. The disadvantage is that separate campaigns may repeat the same steps. The system compensates for this by letting users share stored data and groups of nodes across campaigns.

CampaignRunner reads data directly from external sources. These may be operational systems or an existing data warehouse or customer database. The system connects through native drivers to major relational databases including Oracle, SQL Server and DB2 UDB, which allows it to enhance performance by taking full advantage of special features built into those systems. It can also connect to XML, flat files and other sources. CampaignRunner extracts data from the original source and then does subsequent processing in work tables, which may be in the same relational database or a file. But the system does as much processing as possible in the original database, combining requirements from multiple nodes into efficient, sophisticated selection queries.

CampaignRunner can compare and merge data from different sources to consolidate records from different systems and enhance customer files with descriptive information. It performs basic transformations such as aggregation of values across related records and recoding of values through lookup tables. There are no prebuilt functions for name/address matching, although users could create derived values, such as match keys, for this purpose. But it’s hard to imagine this proving as accurate as specialized matching systems.

Once customer data is assembled, additional nodes can split it into segments. Segmentation rules are written in simple statements, but users can also create complex derived variables and aggregations. By applying such variables and linking nodes in sequence, users can define complex queries without sophisticated programming. The system provides prebuilt functions for ranked, Nth and random selections. Campaigns involving large numbers of cells are simplified with a “multi-split” node that can incorporate up to 100 different cells, each with its own selection statement and each automatically excluding records selected by previous cells. This set of capabilities–complex queries, random samples, large numbers of cells and automatic exclusion of previous selections–pretty much defines a sophisticated campaign manager.

One weakness is that the system currently lacks an automated scheduler. This makes it harder to execute campaigns that involve multiple promotions over time. It also indirectly limits campaigns driven by external events, such as visits to a Web site or responses to a previous promotion, since these are typically implemented by automatically checking for such events at regular intervals. The vendor promises an advanced scheduler with these and other capabilities by the end of the 2005.

CampaignRunner also lacks the variety of auxiliary features offered by some vendors, such as predictive modeling, budgeting and financial analysis, promotion calendars, project management, and real time integration with touchpoints such as Web sites and call centers. These are not essential to campaign management itself, although they can be important in some situations.

The system does provide a tiny bit of content management, insofar as it keeps a list of offer items. These can be grouped into offers which are in turn associated with campaign cells. Users can also assign a source code to each cell. Segments can be output in pretty much any database or file format, including XML and HTML. Personalized letters or email messages can also be generated from user-created templates. Selections can be stored in promotion history tables and posted back to source databases.

CampaignRunner provides modest reporting, including user-defined columnar, multi-level and crosstab reports. More impressive than the reports themselves is that they can be generated automatically at any step in the process flow. This allows users to analyze intermediate results in detail. The system automatically displays the number of records selected by each node and lets the user browse the underlying records themselves.

The user interface for CampaignRunner is quite straightforward. Process flows are built by selecting nodes from a palette and linking them. Nodes change color to indicate their status: red for not yet configured; green for ready to run; and blue for successfully completed. Node configuration, such as specifying data elements and segmentation rules, is reasonably intuitive. When a node is changed, the system recognizes that subsequent nodes in the process need to be rerun. Because intermediate results are stored in work tables, it can rerun the process starting from the changed node rather than from the very beginning. This can save considerable time.

Pricing of CampaignRunner is based on the numbers of databases, servers and users, but not the size of the databases themselves. A single database and three users starts around $60,000, while an enterprise-wide license with unlimited everything costs about $160,000. The system is written in C++ and can be compiled to run on Windows, Unix and Linux workstations and servers.

CampaignRunner was originally introduced in 1998 and has several current installations.

* * *

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics. He can be reached at draab@raabassociates.com.

Leave a Reply

You must be logged in to post a comment.