|
September 2001
BUSINESS RULES
The Fall and Rise of Workflow Software
by Bruce Silver
Workflow is coming back. Perhaps you thought it never went away. Over the last several years, workflow has become an increasingly ubiquitous part of the IT landscape. In fact, in this year's Gartner survey of technology adoption for AIIM International, virtually every company polled claimed to be implementing workflow.
But as an independent category, something you could describe as a distinct software market, workflow for the past couple of years has been dead as a doornail.
For software companies that once defined themselves as "workflow vendors," projections of a booming market for that technology simply never materialized. In spite of its current ubiquity, workflow never spawned an Oracle, Siebel, SAP or BEA Systems - a billion-dollar market "gorilla" that gives a category both a concrete shape and investor appeal.
Instead, each of those giants, along with Microsoft, IBM and countless others, has simply incorporated workflow as a feature of its own flagship application or platform. True, the meager shreds of workflow functionality offered by those products represent but a shadow of the capabilities of dedicated workflow tools, but they have been sufficient to reduce a once-promising software category to a check-off feature.
To be more than just a feature of a packaged application, workflow has always required the scope of the business process it manages to be broader than the scope of a single application. That model worked reasonably well from the mid-'80s until the mid-'90s, linking point applications across the back office. But the rise of true enterprise applications in both the back and front office killed that opportunity, despite the ever-increasing richness and scalability of workflow offerings. While it still can be used to integrate enterprise applications within a broader process framework - adding value in rule-driven approvals, exception handling and deadline management - workflow too often ends up as the tail trying to wag the dog, and enterprise apps don't like that very much. Too often the capabilities of an external workflow engine are overkill, a technology solution looking for a problem.
So why is there suddenly life in the workflow market? The answer, as always, is e-business.
E-business is a whole new ball game. Today, packaged applications don't cross enterprise boundaries, but all companies are desperately trying to define and automate interenterprise business processes. By definition, e-business today implies application-to-application integration in an automated business process. Suddenly, a business problem exists that is complex enough to justify software dedicated to workflow automation.
Workflow in the e-business space is now called e-process. Vendors go to great lengths to insist e-process is "more than workflow," but all the familiar elements are there - automatic sequencing and routing based on business rules, deadline tracking and escalation, management approvals and the ability to change process flow without recoding individual steps.
Two things are different between e-process and the traditional world of workflow. One is the technical architecture. Instead of client/server, e-process engines are built using the new J2EE Web application server infrastructure and XML, the architectural building blocks of e-commerce applications. The new architecture represents a steep learning curve for traditional workflow vendors, forcing today's e-process buyer to choose between e-business infrastructure providers, strong in Java and XML but new to business process management, and born-again workflow vendors who know the process side but are playing catch-up on architecture.
A second difference is the integration of automated "straight-through" processing with exception processing in a single management framework. While in principle capable of handling both, workflow historically has been applied only to that fraction of transactions that required some human intervention - data entry, management approvals or case management. Thus, for example, a health insurance company would have one system for processing paper claims from doctors and a different one for processing EDI claims from hospitals. Only the first would be built on workflow; the latter would use enterprise application integration (EAI) middleware. E-process, at least in the form offered by the former workflow providers, can manage both in a single framework. The new buzz word is "total business integration."
A good example of the shift from workflow to e-process is FileNet's recent announcement of Brightspire, an e-process framework for collaborative commerce applications. FileNet practically invented workflow in the mid-'80s. But by the late '90s, as the company repositioned itself, it no longer even mentioned workflow. Now, as the heart of Brightspire, workflow - sorry, "e-process" - is back at FileNet with a vengeance.
To compete with Web application infrastructure players like Tibco and WebMethods, FileNet is bringing a hard-learned lesson from the workflow wars: Customers today really don't want another toolkit. They want solutions they can deploy quickly, and the flexibility to adapt the underlying business process to changing conditions. So Brightspire will be sold mainly in the form of partner-built applications, not as more middleware plumbing.
The tough times may be over for workflow software; reborn as e-process, it's coming back.
Bruce Silver (brsilver@earthlink.net)
is president of Bruce Silver
Associates, Aptos, CA, 831-685-8803. Reports are available at
www.brsilver.com.
|