Sales of large scale database marketing software took a sharp jump last year, reaching 231 net new installs compared with 177 the year before and 147 two years prior. These figures, compiled from the new edition of the Raab Associates Guide to Database Marketing Systems, suggest that the long-predicted expansion in sales of database marketing systems is finally be under way.

But investors hoping to find the next Amazon.com should look elsewhere. A closer examination of the statistics indicates this market is not yet ready for a dramatic take-off–and may never be.

The first reason for caution is the growth rate itself. Last year’s increase of 31% is impressive, and higher than 20% increase recorded the year before. But it is far from the triple digit growth rates experienced by truly explosive markets like the Internet. In fact, when measured against the installed base of the industry, the growth rate actually decreased–from 40% in 1997 to 30% in 1998. And in terms of dollars, the industry is still minuscule: the 231 sales almost certainly represents less than $100 million in software revenue. Even when services and maintenance on older systems are added, software-related revenues are probably under the $200 million mark.

A second caution signal is that no clear market share leader has emerged–something that is necessary so conservative buyers know which band wagon is safe to jump onto. The fastest-selling database marketing product last year accounted for just 13% of the 231 net new installations, which is not appreciably higher than the 10% to 8% shares of the four next-best-selling products. Nor was there any product with even a 20% share of the total installed base.

This paralyzing fragmentation is likely to continue. Many systems occupy strong positions in particular industries or geographic regions, or have specialized functions such as integrated statistical modeling or loyalty programs. It will be nearly impossible for a single vendor to beat each of them on their home turf. Moreover, new vendors with still other specialties, and often lower prices, continue to enter the market–a record-breaking dozen new products appeared last year alone. Finally, in a particularly vicious cycle, fragmentation ensures price competition among vendors striving to increase market share, which deprives them of revenues needed to fund product enhancements that might let a clearly superior product emerge. The availability of external funding does mitigate this last problem, however.

Another precondition for market take-off is wide agreement on standard approaches. The marketing database industry might seem to meet this one: certainly the past few years have seen emergence of a common model involving a relational database structured to support real-time analysis, batch campaign selections, and response evaluation. During this period, the share of the chief competing technology, “proprietary” systems using flat or inverted files, has fallen from 69% to 23% of new installs.

But the new relational standard is itself increasingly challenged by systems that are designed to meet the new requirement for online interaction with customers. Several vendors–none of them current market leaders–have introduced data models that facilitate interactions by storing an easily-accessible current customer view along with the detailed historical information common to the standard approach. This view can be accessed to look up information about an individual customer as a transaction is under way, rather than waiting to select groups of customers in overnight batch selections.

This new model effectively rearranges the boundaries among different enterprise systems. In the standard approach, the marketing database serves as both the primary analysis system and an operational tool for selections and promotion history updates. Separate “touchpoint” systems, like call centers, sales automation, service desks, and ATM machines, execute the actual customer interactions. At most, the touchpoints are fed lists of messages to deliver to individual customers when they appear. This model has the virtue of allowing separate data structures for the operational systems, which need quick access to individual records, and the marketing system, which must process records in large groups.

The new model splits the marketing system itself into an analytical piece, used to design marketing policies, and an operational piece, used to execute those policies against individual customers. The operational piece then is tightly integrated with the touchpoint systems themselves. It resolves the conflict inherent in the old model, of running operational and analytical processes against the same structure. This dual purpose has long prevented marketing systems from fitting the mold of a classic data warehouse.

The primary reason to split the analytical and execution functions is to permit marketing involvement in real-time transactions–that is, to allow the marketing system to help guide operational decisions as they are made. This is needed both at traditional customer touchpoints and, especially, on the Internet. But other systems are also involved in making operational decisions: for example, a pricing decision might involve consider inventory levels, demand projections and profit margins as well as customer information. So in the new model, the marketing system is really just one of several “peers” that provide an input to the final decision. This requires a radical change in how the system is conceived and designed. It also something that Internet-oriented campaign managers have been designed to do from the start.

Splitting the two functions also has the useful side effect of permitting use of specialized, non-relational databases for analysis. These might be databases like MarketPulse and AnalytiX, originally developed as full-scale alternatives to standard databases, or newer systems like Broadbase, Digital Archaeology or Sand Technologies Nucleus that were designed specifically for data exploration.

Thus the need for real-time interaction presents a challenge to the current standard data model–and to the market leaders whose systems all use it. But the rise of interactive marketing really poses an even more fundamental threat, because it is feeding the growth of competitors who may eventually supplant database marketing system vendors altogether.

These are the developers of Customer Relationship Management (CRM) systems. While the term CRM has now so fashionable that it is applied to nearly everything, it is primarily associated with vendors like Siebel, Vantive, Clarify and Pivotal. These firms build systems that allow all enterprise touchpoints–order processing, customer service, field sales, etc.–to share one database. This gives everyone in the company a complete picture of all interactions with each customer, ensuring that the customer is treated consistently through all channels.

\CRM systems are large, complicated and expensive–and selling very, very well. This is a market that truly is doubling from one year to the next, and already exceeds $1 billion in annual revenue. It does have a dominant vendor, Siebel, whose 1998 sales were nearly $400 million. In other words, this is a market that dwarfs the database marketing software industry.

And there lies the danger. Today’s CRM systems have relatively primitive database marketing capabilities, which leads the database marketing vendors to modestly offer their own products as the central intelligence to control CRM execution. In fact, most marketing database vendors’ visions of growth are ultimately based on their belief that they can piggyback on the expansion of the CRM industry.

But the CRM vendors are rich, smart and ambitious. They haven’t added superior database marketing because their customers haven’t demanded it–either because the customers are too busy just putting in place the mechanical connections among the CRM components, or because the customers don’t really feel a need for better database marketing than the CRM vendors already provide. Either way, it seems a safe bet that should the demand appear for better database marketing, the CRM vendors will add the functions to their own systems rather than turning to outside products. And once they do, database marketing systems will be hard pressed to maintain their current sales levels, let alone generate continued growth. So there is reason for caution indeed.