Two Ways to Win at Process ImprovementTwo Ways to Win at Process Improvement

The surest way to win consistently in baseball is to field a complete team that combines hitting, pitching, defense and team speed. But investing in a few big bats is simpler, and you can still win a lot of games. The same is true of business process management (BPM), which seeks to improve business performance by optimizing processes from beginning to end. You can go with a complete set of software tools--a BPM suite--to address every facet of process improvement: operational cost and cycle time, compliance, agility and performance visibility. But you can also score big wins simply by modeling, analyzing and measuring performance from a process perspective.

'BPA' Versus 'BPMS'

BPM is a management discipline that looks at the enterprise not as a collection of functional units like sales, manufacturing and customer service, but as a set of business processes that cut across traditional stovepipe boundaries. Out of this one basic idea, two distinct software technologies have evolved.

One, called business process analysis (BPA), provides tools that let business users capture how the current process actually works, describe how an improved process should work and describe the proposed improvement in terms of key performance indicators (KPIs) and other quantitative metrics--without having to know the technical details of the underlying IT. Increasingly, BPA tools are augmenting these analytical modeling capabilities to actually measure the KPIs from the process in operation, a capability called BAM.

Other than specifying "business requirements," BPA imposes no constraints, methodology or tools on IT's process implementation. But by simply embracing end-to-end "process thinking," BPA represents a radical change from the functional perspective underlying traditional business practices in everything from enterprise applications to organizational charts. Looking across these boundaries, BPA helps align IT with strategic goals by linking technical innovation with quantitative, business-defined measures of success. And BPA helps foster standardization and agility through management dashboards that make end-to-end performance metrics visible in real time.

The other software technology, called a BPM suite (BPMS), extends BPA with new tools for building IT process implementations. A BPMS represents the complete team approach. While modeling simply describes a proposed process improvement, a BPMS actually implements the improvement by executing the model, automating the workflow, integrating disparate business systems and enforcing the business rules.

A BPMS includes BPA and BAM but inserts a process execution environment in between them. As such, it represents a new way to design and build IT solutions based on a unified set of tools and middleware from a single supplier rather than a best-of-breed component stack. A BPMS also implies a kind of "top-down" methodology in which the executable design is directly driven from the model. Although it can promise a broader set of benefits, a BPMS poses a bigger challenge to traditional IT methods.

Capturing 'As-Is,' Designing 'To-Be'

So do you need a BPMS to "do BPM"? Not necessarily. Although some of BPM's familiar promises--faster cycle times, increased compliance with policies and rules, or more agile IT implementation--are best realized with a BPMS, other benefits, such as business innovation and performance visibility, have been achieved simply by adopting cross-functional "process thinking" in combination with BPA.

Some practitioners maintain that really big cost reductions are more readily achievable through BPA alone than with a BPMS. For example, one consultant working with the U.S. Defense Department is using BPA to help the various armed services move to common weapons systems. Executing the underlying processes is too complex to even think about. "Besides," says this source, who wished to remain anonymous, "why would DoD want to save millions with a BPMS when it can save billions with modeling alone?"

Most companies probably don't have business innovation opportunities measured in billions of dollars. But the example points to the biggest benefit of analytical modeling: getting everyone on the same page. As companies add global operations and grow through mergers and acquisition, it is common to find the same process performed differently in each unit. Although the potential savings from consolidating information and information systems can be huge, these efficiencies are impossible to achieve without a shared understanding of the processes that use those systems--their steps, resources and management metrics--expressed in a common vocabulary and reference model. Process modeling provides that. (Read the "Field Report" on page 33 to learn how Austin Energy used modeling to develop composite-app-based process improvements.)

For Volvo AB, for example, to effectively consolidate IT assets and prioritize future IT investment, the company first had to create a common Order-to-Delivery reference model across its Volvo, Mack and Renault truck brands. Using the ProVision modeling tool from Proforma, Volvo began by mapping processes in the current Order-to-Delivery chain to steps defined in the Supply Chain Council's Supply-Chain Operations reference (SCOR) model, specifying attributes such as inputs and outputs, performance metrics and best practices. This mapping simply captured and documented critical differences in as-is processes between brands in terms of use of SCOR best practices, objectives and performance metrics, and opportunities for improvement. By using a common format for describing current practice, the Volvo team was able to model a proposed common reference model for the three brands. Even without improving any of the underlying processes, modeling enabled Volvo to begin managing IT effectively at the enterprise level rather than at the brand or factory level.

Optimizing the Process

Capturing existing processes and describing proposed "to-be" processes--their steps, resources, goals and performance metrics--in a standard format for shared understanding is the first essential function of analytical modeling tools. The second is estimating process performance through simulation analysis and optimizing the model in a variety of scenarios for lowest cost, shortest average cycle time or conformance to some other KPI.

Simulation analysis requires applying parameters to the model in terms of estimated resource and activity costs, milestone-based benefits and the relative likelihood a process instance will follow each path in the flow. Based on such parameters, the simulation engine in a process analysis tool can project mean and maximum cycle times, costs and values of other defined metrics.

Business analysts, not IT, typically perform modeling and simulation. These steps demand business process knowledge, but they don't require technical understanding of the underlying IT systems implementing process activities. Advanced business process analysis tools like Proforma's ProVision or IDS Scheer's ARIS bring additional richness and sophistication to the model and simulation analysis, supporting external events, business rules and reference models such as SCOR, IT infrastructure library (ITIL) and Sarbanes-Oxley.

Although such dedicated process analysis tools are extremely powerful, the vast majority of process diagrams in existence are Microsoft Visio drawings. Visio is inexpensive and, unlike the high-end tools, requires no training and imposes no diagramming standards or methodology. That very freedom limits Visio's ability to make models sharable across the enterprise, support simulation analysis or even document process requirements with sufficient precision to enable IT implementation. To address the ubiquity of Visio process diagrams, most dedicated modeling tools let you import Visio diagrams, mapping Visio shapes to the shapes and semantics of the modeling tool. In addition, products like Byzio from Zynium provide a similar mapping and import to a variety of BPMS offerings.

Measuring Actual Results

A third function found in some BPA tools is business activity monitoring (BAM). Where simulation analysis projects estimated values of KPIs, BAM actually measures them from running IT systems, aggregating information from a variety of sources, displaying them graphically in management dashboards and triggering alerts to process owners when KPIs deviate from their target values. In some process analysis tools, simulation results are displayed in the same dashboard used for BAM. This facilitates comparison of the modeled and actual performance results, as well as feedback of actual parameter values to refine the simulation model.

This BPA-style BAM is different than the BAM provided by a BPMS, which offers a simpler way to collect performance data and a platform for triggered escalation actions. For example, without a BPMS, BAM requires its own integration adapters, data mappings and other middleware to collect and correlate performance data directly from backend systems. A BPMS includes these capabilities. Even without a BPMS, BAM can only post alerts to a management dashboard, while BPMS gives BAM a ready platform for responding directly to the process with automated escalations.

But even without a BPMS, BAM enables business process analysis to fulfill two of BPM's key promises: business-IT alignment and performance visibility. This appeals to IT organizations that resist the idea of using an integrated suite of tools from a single vendor or BPMS's top-down, business-driven style of solution implementation.

When A BPMS Shines

Many processes exist for which business process analysis, by itself, can't compare with the results achieved with a BPMS. In particular, human-centric processes--those in which process performance is measured in terms of the quality, speed or cost of human work--depend on BPMS features such as workflow automation, business rule management and automated exception handling to optimize business performance. Certainly, IT can build comparable process implementations without a BPMS, but the latter provides a model-driven, zero-code approach. A BPMS unifies human tasks, application integration and business rules out of the box, making process implementations much easier, faster and less costly to build and maintain.

BPM suites also include BPA tools, but these bundled modeling and simulation offerings may lack all the bells and whistles provided by standalone BPA vendors. Some BPMS vendors, like Savvion, have made the modeling and simulation component free and the focal point of their market education efforts. Others, such as webMethods, use their process analysis tools as part of service engagements based on a methodology for discovering processes with the maximum potential ROI from BPM.

Whether or not you end up using a BPMS to build and execute process implementations, the first step has to be process modeling and analysis. Even if that's the only step you take, you can achieve real bottom-line benefits through business process improvement.

Bruce Silver [[email protected]] is an independent industry analyst and author of the blog BPMS Watch. Engage in the discussion.

Take Action

Spot overlaps and inconsistencies. As companies grow through mergers and acquisitions or global expansion, redundant systems and variations on the same business processes are inevitable. Use a process modeling tool to capture "as-is" processes and spot overlapping technologies, data sources and work steps.

Get on the same page. Model proposed "to-be" processes, creating a shared vocabulary and forging broad agreement across IT and business units on process steps, resources and management metrics.

Estimate process performance. Simulate and optimize processes in the modeling tool by setting parameters for resource and activity costs, milestone-based benefits, mean and maximum cycle times and other metrics that serve as KPIs.

Measure actual performance. Use BAM to measure and aggregate information from running IT systems, displaying actual process results in a management dashboard and triggering alerts when KPIs deviate from target values.

Eliminate manual work steps. You can't eliminate all human interaction, but you can optimize process performance with BPM suite features such as workflow for exception handling or business rule management for complex, high-volume decisioning.

Field Report

Austin Energy isn't using a business process management (BPM) suite per se, but it is realizing business process improvements as part of its enterprisewide service-oriented architecture (SOA) initiative.

Two years in the making, the SOA initiative is helping the $1 billion electric utility integrate and leverage disparate legacy systems. It's also helping the business adapt in the face of deregulation and new competition driven by new energy storage and low-cost power generation technologies.

"The energy business is rapidly changing, so the company wanted to free up capital to fund new business models," says Andres Carvallo, CIO. "That meant 'right sizing' the current IT contracts and maintenance agreements to free up dollars to build an SOA driven 100 percent by the business processes."

On its way to building the new architecture -- including an enterprise service bus (ESB) and enterprise portal built on the IBM WebSphere application server and middleware -- the utility decided to tackle business processes very tactically, with well-defined requirements and desired benefits. Work on the first of those improvements began last November, when the company targeted the customer-support process around power outages. Austin Energy used the WebSphere Business Modeler to spot bottlenecks and potential improvements in the existing process, which was supported by a system written in COM objects and Visual Basic. Scalability was the key problem, as the system could only handle about 4,000 calls per day. As the calls came in, the customer service reps couldn't log on fast enough, they experienced lags referencing customer information in the databases and billing systems, and they couldn't create work orders quickly enough.

To speed the process, the utility created a composite application comprising five Web services that link existing systems and databases on a robust SOA infrastructure. "The new application went live on May 3, and the very next day we had a storm and were able to process 20,000 calls," Carvallo says.

Average call times dropped to one-and-a-half minutes from the previous range of three minutes to five minutes. Adding to the savings, work orders were streamlined from a paper-based, batch-oriented system to direct integration with a workforce management and dispatching system.

"We haven't been able to quantify the savings yet, but when you can only get to 4,000 calls -- instead of 20,000 -- you're missing a lot of outage information that can give you a better picture of the source of the problem," Carvallo says. "That means you know right where to send the repair trucks, and that's a big deal, given that every truck roll costs at least $70."

Austin Energy is currently using the Modeler to find ways to speed and automate as many as 72 processes related to turning off and turning on service for customers and dispatching service crews. "Anything that involves mobile support, because the ROIs are very measurable, the benefits are quickly realized and they have a big impact on customer satisfaction," the CIO says.

The utility is also exploring how and whether to integrate customer information systems with an interactive voice response (IVR) system and the enterprise portal. "With personalization based on your account registration information, you could enter a PIN number or account information and see exactly when your power will be turned on or restored," Carvallo explains.

Where does an SOA initiative end and BPM begin? In Austin Energy's case, the literal point of intersection is the business process modeling tool. But from Carvallo's perspective, "if you don't have your infrastructure, data models and integrations in place, you can't do enterprisewide BPM. Maybe you could do one BPM solution at a time, but to me, SOA was the solution we wanted because we were trying to solve a corporatewide challenge."

That said, Carvallo says the crucial difference between SOA and previous architectural approaches is that business processes drive it. "You map to the business process and you rationalize the infrastructure that delivers the functionality that makes that business process possible." --Doug Henschen