The concept of logical processes underpinning business functions should warm the cockles of business-technology executives' hearts. In business, processes determine everything from what happens when an employee is hired to the steps taken when a customer places an order. Some organizations have even patented their processes in custom software that automates the steps involved in building a perfect widget.

The problem is, like code, business processes can become bloated. As a business grows, slavishly clinging to outmoded processes will cut into profitability. Processes can--and should--evolve with the dynamic needs of a company. Automating business processes with a shiny new business-process-management implementation does wonders for simplification and optimization. For example, the discovery phase before putting BPM into operation forces companies to define the steps in each process, often revealing redundancies and inefficiencies. Automation also lets companies change processes rapidly to adapt to business conditions.

The BPM market is growing at a 20% clip each year as a scattered field of more than 100 vendors begins to consolidate. The market's $1 billion in sales in 2004 should reach $3 billion by 2009, according to researcher IDC.

Most BPM suites are process-oriented, sharing information among modeling tools, fat clients, portals, and the process engine through Web services. The advantages of this model are reuse, interoperability, and faster time to deployment. In the recessionary years of 2001 and 2002, cost cutting and productivity initiatives drove BPM demand. BPM helped speed process cycle times with integration and automation that closed the gaps left by applications such as enterprise resource planning. BPM systems also provided a faster, flexible, less expensive, and business-friendly environment in which to model and change processes.

"The downturn proved that BPM saves money," says Gartner analyst Jim Sinur, pointing to a 2004 survey of 50 people implementing BPM in which 95% said their projects had been a success. Respondents reported an average 15% rate of return, and 55% had returns in the $100,000 to $500,000 range on each project.

While the first waves of BPM adoption were about automation and integration, the focus today is increasingly on regulatory compliance, business and application agility, and optimization. Driven by mandates such as the Sarbanes-Oxley Act and Basel II, companies are turning to BPM to enforce policies and procedures in financial reporting and other core processes that are material to company results. This is a natural for BPM because the systems are designed to document and model processes, apply explicit policies and procedures (as enforced by rules), execute in a consistent hands-off fashion, and track results.

In the drive for business agility, business-process management has much in common with services-oriented architecture. Both aim for faster response to changing business requirements, starting with compliance but also including mergers and acquisitions and product and service introductions. In fact, 53% of Web-services-integration applications are applied to business processes, according to an Evans Data study. So services-oriented architecture has become a crucial foundation for BPM, supporting rapid assembly and orchestration of process services into larger, end-to-end processes.

Another goal is tying business-process management to business-performance management so that process improvements serve strategic performance goals. If improving product quality is at the top of your agenda, process-management initiatives should focus more on qualitative measures and less on cost reduction and faster cycle times.

To connect processes with performance goals, companies need business-activity monitoring features, including metrics, key performance indicators, executive dashboards, and advanced reporting capabilities. InformationWeek's sister publication Network Computing recently tested nine BPM suites. (For the results, see Business Process Management Suites). All nine products offer basic operational metrics, but many fall short on the promise of easy access to higher-level data and real-time information--they provide access to the data but not good reporting. And few products close the loop on processes by applying operational results to continuous process improvement.

Simulation was a crucial differentiator in Network Computing's tests. Every product has a "play" button that lets users run through the basic process, but only a few offer deeper simulation in which users can project human and systems resource requirements, and fewer still let them start, say, instances of a process and do serious diagnostics and cost planning. More important, it's still nearly impossible in all but a few products to simulate more than one process simultaneously. This is a problem because it's unlikely that people are dedicated to a single process, so it's important to simulate not just one but all processes involving the same people. This limitation makes business-activity monitoring and historical analysis even more important for capacity planning and process optimization.