InformationWeek

September 14, 1998

Decision Support Cycle Unique

By Jacques Surveyer

ith Microsoft's entry into the planning and decision support services market, a legitimate question would be, Is the P&DSS market ready for commodity pricing? The figure on page 319 shows the simplicity and complexity of decision support software. The systems can be seen as a simple pyramid model with online transactional processing and external sources feeding into higher-level planning, budgeting, and decision-making mechanisms. These information pyramids often map quite closely to a hierarchical or hub-and-spoke data warehousing model. What could be so difficult? Build a hierarchy of data warehouses and, like Shoeless Joe, they will come.

Unfortunately, many an enterprise data warehousing project has foundered on just such notions because a good portion of P&DSS is ad hoc, responding to amorphous and evolving needs of problems or projects. Mapping data marts and warehouses to planning processes is not trivial because, in addition to a central core of summary information, P&DSS problems often require different viewpoints, drill-downs, and cross-functional or external information. True, planning and budgeting processes tend to follow a predictable time cycle, but not necessarily a problem agenda. So, in effect, five critical P&DSS factors emerge:

The requirements for information and data tend to grow and recede, shift, and return;
The demands on IT systems and, equally important, interfaces between systems vary over time;
Timeliness and accuracy of data and information is as critical as their presentation;
Historical trends and patterns are vital for insights and building predictive models;
Planning and decision making are political as well as financial/technical processes; knowledge of the local terrain and history is vital.

Responsive To Change
The result is that P&DSS systems have to respond much better to changing sources as well as volumes of data. Thus, unlike in OLTP and operational systems, many a P&DSS project is never completely "done." Thus, P&DSS software and processes tend to be:

Open and highly linkable to a wide range of cross-platform and/or ad hoc data sources;
Flexible and customizable to meet changing requirements;
Easy to use and then reuse (come back to full speed in minutes) after some indefinite layoff in usage;
Scalable upward, to meet sudden, unpredictable, and heavy processing demands for analysis;
Responsive, able to meet speed of thought and what-if demands that must be accomplished in seconds or minutes, not hours; and
Scalable downward, to archived and condensed histories for post-decision review or contingent new needs.

Given that people costs outweigh P&DSS hardware and software costs by at least 10 to 1, and given the lack of skilled IT personnel, best-of-breed P&DSS software or hardware that helps to add 10% to 20% more productivity can quickly add value that more than covers its costs. Likewise, having a P&DSS vendor and/or reseller that has industry-specific experience and that can offer a range of ongoing support, mentoring, consulting, and even turnkey services can be a critical success factor for P&DSS systems.