David Raab Article Archive

The question coming through the phone seemed so simple, so reasonable.

“What more do I get as I pay more for a campaign manager?”

But there was no simple answer. The campaign manager you buy for $100,000 looks a lot like the one you buy for $350,000 or $1,000,000. True, vendors often charge more for larger customer bases or more powerful servers. But comparisons across different vendors for the same size system still can yield differences in the hundreds of thousands of dollars.

Some of the price differences are more apparent than real. In a competitive situation, the more expensive campaign management vendors have sometimes offered their systems for less than half the list price. In an era when venture capitalists or financial markets willingly fund losses of firms in hot markets, vendors need not squeeze the maximum revenue from each sale.

Still, it’s generally accurate that prices for established high-end campaign managers from vendors like Xchange Inc., Prime Response and Recognition Systems (Protagona) start at around $350,000. These are roughly equivalent products: while each has its strengths and weaknesses, they all support the full range of traditional campaign management activities–complex segmentations, multi-step campaigns, sophisticated champion/challenger testing, and detailed response analysis. Today, most also provide some form of email campaigns, near-real-time response and integrated statistical modeling as well.

The real question coming over the phone was: what would my caller give up, and how much would he save, if he didn’t buy a top-tier system? He had four major options.

– generic query tools. Products like Business Objects, Brio and Cognos Impromptu can all be used to generate reasonably complex queries, which are the true core capability of a conventional campaign manager. After all, the purpose of these systems is ultimately to generate lists, which are nothing more than sets of records selected by a query. Although marketers have traditionally wanted their campaign managers to generate the sorts of complex, ad hoc queries that tie SQL in knots, the reality is that most of today’s high-end campaign management systems do little better at this than generic SQL-based query tools. (An earlier generation of campaign managers, using proprietary databases and non-SQL query languages, were much better at complex queries, but the market chose openness over query power.) Marketers have found that they can live with the compromises needed to execute complex selections within the confines of standard SQL, such as precalculating key aggregates and splitting a query into several steps. Even random selections, the one non-SQL function that nearly all high-end campaign managers have made a point of providing, can be reasonably simulated in SQL-only systems through tricks such as assigning a random number to each customer or selecting on quasi-random values such as the last digit of a Zip code.

So what would a marketer give up by using a generic query tool alone? Two things, really. The first is the set of functions involved in defining and executed multi-segment, multi-step campaigns. A standard query tool executes one query at a time: for a marketing campaign that might have dozens, hundreds or even thousands of segments, this means at least one query per segment. The time and computer resources needed to execute each of these against the underlying marketing database may be unacceptable; even if not, the human effort needed to ensure each segment definition is correct would be impractical. A sophisticated user might make things more efficient through clever use of tags and temporary tables, but most marketers lack the inclination or skills to do this. A high-end campaign manager, by contrast, provides a marketer-friendly interface that lets the user specify how the segments relate to each other, and then figures out for itself how to execute this plan as efficiently as possible. (Truth be told, some of the high-end systems are considerably more efficient than others. But all at least offer a simple interface that removes from the marketer the burden of trying.)

The second function missing from a simple query tool is response analysis. All campaign managers provide automated features to store a customer’s promotion history, match promotions against later responses, and report on the results. Again, the campaign managers vary widely in the details of how they do this, but all of them include a framework to manage the process. By contrast, the buyer of a general purpose query tool would need to design and implement this process from scratch. Even more than complex segmentations, this is something that clearly must be done by IT professionals.

The cost of building a response analysis system must be counted against any savings the result from purchasing a generic query tool. The savings themselves are likely to be substantial: a standard query tool might cost anything from $1,000 to $10,000 per user, with prices tending to the lower end of that range. Since most campaign management systems have only a handful of real users–the marketing planners and analysts who set up and monitor complex campaigns–many firms could buy the necessary software for well under $50,000. For companies with limited needs for campaign complexity and response analysis, this could yield a considerable savings even after the higher development and operating costs are taken into account.

– sales force automation systems. These systems are designed primarily to manage contacts between individual customers and salespeople, in person or over the phone. However, nearly all can generate lists for outbound mail and telephone campaigns and evaluate campaign results. Thus, they provide at least a rudimentary version of conventional campaign management systems. And, being aimed at a set of notoriously tech-hostile users, their interfaces are the easiest available.

Sales automation systems diverge from high-end campaign managers on two critical dimensions. The first is campaign complexity. Most of these products limit each campaign to a single segment. They further limit the complexity of the query that defines this segment to the subset of SQL functions that can be easily represented on a fill-in-the-blanks form. This generally means, for example, no ‘or’ relationships among fields and no calculations. Response definition is also typically limited to a source code captured with a reply: more complicated tagging, based on queries that search a transaction database for specified behaviors, would be custom-coded outside of the system. On the other hand, many sales automation systems do let users set up multi-step campaign sequences, such as an outbound telephone call followed by one or more personal and direct mail contacts.

The second major difference between conventional campaign management systems and sales automation products is the nature of the underlying customer database. Conventional campaign management tools usually run on a denormalized structure designed to handle large-scale analytical queries against the entire file. Even when the campaign manager is adapted to support real-time interactions, this is usually accomplished by adding a subsidiary profile table rather than restructuring the main database to handle online transactions. By contrast, sales automation systems usually have complex normalized data structures, both because they are intended to handle transactions involving one customer at a time and because they are typically designed for business-to-business marketing situations where there are multiple layers of data about companies, locations, individuals and projects. The richness of such a data model is potentially a benefit for conventional campaigns, particularly in business-to-business. But large-scale analytical queries may be painfully slow against a complex, normalized structure.

The simplest sales automation tools, basic contact managers like ACT!, can be purchased for a few hundred dollars per user. But few marketers would find these powerful enough to be of value. More realistic candidates for campaign management are mid-tier products like SalesLogix or high-end products like Siebel. The prices of these are comparable to standard query tools, also in the $1,000 to $10,000 range per user, with most running somewhere from $1,500 to $6,000. There is often a minimum number of users or a central server component that brings the minimum price to $30,000 to $50,000. The result for installations with a handful of users would be a higher software cost than a solution based on standard query tools. Implementation costs might be lower for sales automation systems, since they include some functions that would need to be built from scratch for a query tool. But most marketers would probably need to simplify the sales automation system’s standard data model, a task that could easily eat up any savings in development. So, over all, the sales automation-based solution would probably bemore costly.

– marketing automation systems. This class of system, including products like MarketFirst, Annuncio and Revenio, is built primarily for marketers rather than sales people. The main product of this difference is the underlying data structure, which leans more towards the denormalized, analytical structures of conventional marketing systems than the normalized, transaction-oriented structures of sales automation. In terms of campaign management functionality, marketing automation systems are actually quite similar to sales automation systems: they too offer multi-step campaigns and simple response analysis, but often do not build complex multi-segment campaigns. Perhaps half include random sampling for champion/challenger tests. It’s particularly dangerous to generalize about this segment, which contains a wide variety of systems that are evolving rapidly–and not necessarily all in the same direction. But, compared with conventional campaign managers, many products in this group have more of a business-to-business orientation, offer more integrated Web site and email components, provide more administrative functions such as project management and budgeting support, and are designed for somewhat smaller marketing databases. Some products that fall roughly within this group, notably E.piphany and Broadbase, offer a slightly different mix of features that include tightly integrated analytics, extensive data preparation and reasonably powerful campaign management, spiced with a well concealed dash of proprietary database technology.

Like conventional campaign managers, marketing automation systems are usually priced according to the database size rather than number of users. Minimum prices are generally in the $100,000 to $250,000 range and rise less steeply with volume than conventional campaign management systems. Except for the most demanding campaign management users–people with very large consumer databases who employ sophisticated direct marketing techniques–many marketers will indeed find marketing automation products an attractive alternative to high-end campaign management systems.

– less known campaign managers. Although Xchange, Prime Response and Recognition Systems own the major brands among high-end campaign managers, there are other systems that offer competitive products. Notable contenders with proven scalability and entry price points as low as $100,000 include Decision Software (TopDog) and Unica Technologies (Affinium). A newer product, dbMastery, offers impressive functionality against 500,000 records for $60,000. But dbMastery relies on index-intensive databases like FoxPro or Sybase IQ, and may not easily transfer to the mainstream databases like Oracle and SQL Server that are preferred by most large 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.