Column 2

I missed Justin Brunt’s webinar a couple of weeks ago on BPMN, so I stopped by his session today to see it in person. There’s a lot about BPMN elsewhere on my blog, including my recent BPM Think Tank coverage and the BPMN category, so I’m not going to go into too much detail about what BPMN actually is; in short, BPMN is the business process modelling notation, namely the graphical representation of process maps and some of the associated metadata, intended to be used by both business and technical users as a common vocabulary for describing processes.

Justin walked through all the BPMN object types and how they’re used, plus some multi-step examples to show how they all fit together.

Although there’s a lot more complexity that can be explored, he only had 30 minutes so this was a quick overview. Bruce Silver has a 90-minute workshop on BPMN tomorrow morning that will go into considerably more detail.

Michele Cantara is holding the session on the BPMS market, including key players and trends. I can tell that the straw poll that Daryl Plummer did yesterday is correct in terms of there being more business than IT people at the conference, since this session seems to be shockingly poorly attended. Maybe it’s just that this is the first session after lunch.

Gartner is now pushing the idea of integrated composition environments (ICE) as an expansion on a BPM suite: they position BPMS as one step on the way to an ICE.

Cantara shows the current Gartner representation of BPMS as a set of functionality that is available as a “single product experience”: process component registry/repository in the middle, surrounded by process execution and state management engine, model drive development environment, document and content mgt, user and group collaboration, system connectivity, business event BI and activity management, inline and offline simulation and optimization, business rules management, and systems management and administration.

She sees the BPMS market as having started in 2005, since prior to that, no vendor had all of this required functionality; in 2006, there were 19 of them (including a few that I have never thought of in this space or have never heard of, such as Graham Technology), and there’s another 5 as candidates for the 2007 BPMS market share analysis. The prediction is that 4 or more of these will be acquired by platform or application vendors within the next two years, so will drop out of the BPMS market — a bit of a weird statement considering that IBM and Oracle, which are clearly platform vendors, are still here and are considered the two largest BPMS vendors: why wouldn’t the acquiring vendors just be added to this list if they acquire the BPMS vendor?

There’s some pretty interesting conclusions to be drawn from a chart that she showed of the top 10 BPMS vendors and their comparative 2005 and 2006 market share: many of the vendors had their market share stay the same or even reduce, even though they increased their revenues, since the size of the market grew. Since IBM and Oracle, shown as the top two vendors for 2006 and non-existent for 2005, really skew the numbers: I suspect that they didn’t build that business in one year, but that Gartner reclassified a big chunk of what they do as BPMS and therefore “increased” the market size without it actually increasing all that much. This comes back in a later slide that shows a projection of the market size based on the 2006 numbers: since the market “grew” by 69% between 2005 and 2006, which I think is based primarily on Gartner reclassifying existing business rather than actual growth, another 40% is expected for 2007, then 20% year-over-year until 2011. Short of a whole new bunch of recategorization, I just can’t see the BPMS market tripling from $1.7B to $5.1B (which is the effect of those growth rates compounded over time) over that five year period. The overarching category of “middleware” (PPMW = portals, process and middleware) that includes BPMS is predicted to only double during that time, which looks conservative in comparison.

She showed another interesting chart that plotted business user-driven versus IT driven control against infrequent versus continuous change. BPMS’ fit best in business user-driven, continuously-changing processes, but may also be applied to any business user-driven process and any continuously-changing processes even though less integrated pure-play BPM products may also fit in these latter situations. For IT-driven processes that change infrequently, a BPMS is likely overkill.

She also looked at the four main BPMS buying patterns: automating a specific process, continuous process improvement, moving to SOA, and business transformation. Each of these has different types of buyers and different goals, which can mean quite different requirements for a BPMS depending on what’s driving the original purchase. Market drivers (e.g., compliance) and inhibitors (e.g., market share consolidation) also impact purchasing patterns.

Vendors that can’t play the entire BPMS functionality game that Gartner has defined can take a few different routes. They could concentrate on some best-of-breed functionality that they offer, and partner for the remaining functionality (which as recently as a year ago, was Gartner’s definition of a BPMS). They could focus on specific process-based applications for a small number of verticals rather than a cross-industry horizontal suite. They could focus on an integrated service environment (ISE) with service assembly and some degree of development in a model-driven approach. Lastly, they could focus on a business process platform (BPP) approach.  Cantara discussed the different situations in which you might choose one of these types rather than a full BPMS, and showed some indicators for telling whether a BPM vendors is moving in one (or more) of these directions; in some cases, the distinction between these four routes are pretty vague.

Her final recommendations really position BPMS as a development platform, which is more accurate than not, but not necessarily the message that the BPMS vendors give to the market. She also pointed out the critical role of round-tripping capabilities between modelling and execution environments.

I thought that they’d have a new BPMS magic quadrant out by now, but apparently it’s due next month, so stay tuned to Gartner for that.

The next session was on measuring processes in real time, namely business activity monitoring (BAM), and how it needs to be considered up front as processes are being design and implemented.

Gassman started off with a few definitions — BAM, real-time BI, operational BI, and process-driven BI — with some pretty fuzzy distinctions between some of these, especially in these days of converging functionality in the BI products. He then defined the goals of BAM: to monitor key objectives, anticipate operational risks, and reduce latency between events and actions. From an implementation standpoint, BAM is typically a real-time dashboard that’s integrated with BPM in some way and provides alerts in the context of the processes within the BPMS, but also links to more traditional BI functionality such as predictive and historical reporting. He also makes a distinction between more passive data displays that require someone to be looking at a dashboard in order to detect the condition, and sending an alert when a specific threshold condition is reached.

He talked about a number of different real-time analytic techniques, including process monitoring, logistics optimization, situational awareness, complex event processing, track and trace, and anticipate and react. Of all of these, CEP is the up-and-coming new technique (to be covered in the Event Processing summit that follows on after this BPM summit for the remainder of this week) that uses pattern recognition and matching, whereas the others are based on pre-defined metrics. Looking at it along another dimension, anticipate and react is a predictive technique that uses past and current data to predict the future state, whereas the others primarily display data or events that have already occurred.

Without monitoring, it’s difficult to detect when something has gone wrong in a business process; with BAM, not only can specific business questions be answered based on measurements, but analytical techniques can be applied to detect correlations, report on impacts, detect root causes, and make some predictions. This, in turn, feeds into the broader scope of corporate performance management.

He went on to discuss the synergy between BAM and BPM, and how BI (which he considers to be a different type of functionality) can be tossed into the mix to provide operational decision-making and even trigger new processes, on top of the “awareness” that comes from BAM. Although BAM and BPM are a natural fit, BAM doesn’t just come from the BPM vendors and isn’t just for BPM: some BAM tools are focussed on or part of business/enterprise software and BI suites. Having BAM integrated into the BPMS has some advantages however; the monitoring can be modelled right along side the processes, and drill-downs from the dashboards can go directly back to executing processes. However, many of the BAM products that are part of a BPMS are less functional than their general-purpose counterparts, and may be limited to monitoring just the business processes and not the larger business context. Because of that, Gassman’s final recommendation is to look for a BAM product that can be integrated with a BPMS but can also run standalone.

Yvonne Genovese opened the day with a discussion on business applications through 2010 and the major changes impacting process environments. Her main focus is on enterprise applications, not BPM, but these applications have a huge impact on process management in any organization.

Her first strategic planning assumption: “By 2009, all leading application platforms will support design and deployment of events, complex events and services in a common, integrated context enabling users to create their own BPP [business process platform].” She followed this up later in the session to discuss how the enterprise application vendors will tend to try to increase the footprint of their application with customer organizations by expanding out to use their own middleware and other platform pieces.

No surprise, she also sees enabling process innovation and agility plus incorporating information into the processes will be a key for success: this is the intersection of BI and BPM that’s starting to occur more often today.

She discussed the problems today with fulfilling the dream of using services to build business applications: in reality, many of the services don’t match up all the well, and it’s not just plug and play as the application vendors might have you believe but still requires some amount of integration. However, by 2010, she sees the application environment becoming more service oriented such that services can be more universally used with less concern about the specific architecture of the services. That allows some parts of packaged applications to be more easily replaced by composite applications or outsourcing with much less custom integration work than is now required. For a process environment, this means more flexibility in sourcing processes or portions of processes, and the packaging and pricing will reorient towards operational processes.

As this happens, control will start to shift from vendors — from the environment of a primary vendor ecosystem — to the customers, who create their own ecosystem by selecting middleware, process components, SaaS, outsourcing and other building blocks. Many of these decisions also shift the control from IT to business within an organization, but it’s very necessary to distinguish between the processes that provide an organization with their competitive differentiation when selecting what to outsource or even move to SaaS. For example, SAP is announcing this week that they’ll have an entire SaaS platform for customers to use, but it’s not as functional as the full on-premise suite that you would buy from them directly, nor will it be integrated or customized for your environment to the same degree.

Getting back to the recurring theme of people within the process that I’m seeing this week, she also stated that the focus will move from process efficiency to arming the information worker. In fact, she states that by 2010, more than 50% of collaboration and user productivity interactions will be integrated with process technologies such as BPM — this addresses not just the transactional human interactions that occur in many automated processes now, but also all those messy, event-driven, less structured human interactions that are almost purely manual now.

Genovese went through three different process governance models — safe and steady control, cautiously dynamic, and aggressively predictive — and discussed the pros and cons of each as well as their basic characteristics.

We then arrived at this week’s first explicit definition of what Gartner means by a business process platform:

The business process platform is a combined IT and business model that enables enterprises to accommodate rapid but controlled business process change through the use of integrated process composition technologies and the delivery of reusable business process components and their management.

Whereas at the conference in February, they were saying that the large vendors would be creating BPPs and organizations would be buying them, it now sounds like they think that customers will be rolling their own BPP from best-of-breed components that best fit their environment and business needs, and assembling a virtual business services repository. In this context, it’s especially important to build or buy services with reuse in mind.

Genovese finished with a review of some of the large vendor’s BPP architectures, including a view of their BPP stack, and a summary of their strategies, strengths and challenges: Oracle, SAP, Microsoft and IBM. Her final recommendation for end-user organizations brought home why it’s important to understand what these vendors offer:

This era of change will require many users to re-platform all or some of their process environments; therefore, understand how the current business applications changes affect your environment now.

Good advice.

I can’t believe how quickly today has passed: it’s already time for the closing keynote of the day with Daryl Plummer. First, however, he showed us six predictions and asked us to vote to see which will be covered in the research session on Wednesday:

• By 2013, your executives will no longer have control over more than 40% of the people they depend on.
• Business people will become adept more quickly at business-IT alignment than will IT professionals, causing many It professionals to be shut out of BPM leadership.
• By 2010, job opportunities for IT professionals lacking business expertise will shrink by 30%.
• Packaged business applications vendors with implicit process models in their solutions will be relegates to commodity status by 2010.
• Through 2010, SOA, SaaS, BPO, open source, business application implementation projects that don’t make process integrity an integral part of the implementation will fail. (The SOA SaaS, BPO, open sources on the front end of this sentence didn’t make much sense)
• By 2009, less than 10% of BPM project revenue will flow to offshore services vendors.

The real focus of his keynote, however was on how BPM tames the SOA beast and the related keys to success. Plummer, who’ve I’ve always seen as an SOA guy in the past, has really come around to the BPM party line, stating that SOA is “one mechanism that makes BPM easier to do consistently well” rather than “the foundation of all life as we know it”. Maybe it’s due to the survey that they did finding that CIOs’ #1 priority is now improving business processes.

He started out with a wonderfully amusing description of SOA and web services in business terms; I’d be surprised if there was anyone in the room, no matter how non-technical, who didn’t understand it when he was finished. He then moved on to the three primary issues of the session: the relationship between BPM and SOA, how BPM can be used to govern and evolve an SOA, and what technologies, vendors and practices are most effective for building SOA success. My favourite quote from his presentation: “SOA can have an unexpected impact on people; BPM can have an unexpected impact on systems” — what an understatement!

In covering the evolution from distributed computing to SOA, Plummer pointed out that we’ve moved from integration technologies (RPC, CORBA, etc.) to the interoperability enabled by web services; on the BPM side, we’ve evolved from automation to process centricity. Interestingly, in the evolution of SOA we see where the need for BPM arises, and in the evolution of BPM we see where the needs for SOA arises. He brings it all together under the umbrella of enterprise architecture, where BPM provides the business viewpoint and SOA provides the technology viewpoint, all within the context of business strategy and supported by the underlying technology infrastructure.

He moved on to discuss critical end-to-end processes: “processes that provide a unified function but depend heavily on multiple subprocesses, none of which is intended solely for the purpose of satisfying the E2E process.” The subprocesses on which they rely are the siloed functional processes, but it’s the end-to-end process that makes a difference in the organization’s success.

Of course, when you start talking about SOA and service assembly into processes, the natural progression is to look at compositions instead of monolithic applications, and from there to what Gartner refers to as service-oriented development of applications (SODA): designing and developing components for reuse and dynamic binding. This, in turn, requires the right kind of process-centric people to bridge between the worlds of BPM and SOA using tools such as a BPMS for iterative process implementation and improvement.

Another of the vendor sessions, and I’m sitting in on the BEA session, at least for Jesper’s half of the talk (I’m still torn on whether to stick around for the second half or scoot over to the Savvion user panel).

BEA is really realigning their message to put equal weight on their Enterprise 2.0 platform (which I covered extensively at their conference earlier this year) along with BPM and SOA.

Jesper started with a bit of a BPM market review, somewhat unnecessary given the information from Gartner that we’ve seen so far today, but quickly moved on to the most common challenges that they’ve determined based on a recent survey with their customers: inflated expectations (which relates directly to the position of BPMS as shown on the BPM hype cycle that Janelle Hill showed in the previous talk), organizational barriers, deployment complexity, user acceptance, and lack of skilled business analysts (to which Hill also referred).

He covered some best practices such as building a centre of excellence and using an iterative/agile methods, and discussed a matrix for prioritizing BPM implementation projects based on the cross product of their complexity and their process impact: start with the simplest process that has the least impact, then move along the direction of increasing process impact, then return to low-impact, high-complexity processes before finishing with high complexity and process impact. He showed a sample of a prioritization matrix with various processes for an organization plotted out by complexity and impact, which would offer a good guideline for planning the implementation of BPM projects.

In their lifecycle assessment survey (I believe that this is the one that they have online on their website), they’ve seen that there’s a measurable impact of SOA on the success and speed of BPM projects. This isn’t surprising, but it’s good to see some empirical data to back it up.

As a follow-on from some of the earlier comments from today, Jesper also sees that the future of BPM will move beyond the standard type of transactional flows that are common today and into more collaborative interactions and decision-making that involve ambiguity and require tacit knowledge, often as part of the same processes. That goes back to the same message that the people in the process are becoming more of the focus for BPM; it’s my opinion that we’ve done so much in automating tasks and streamlining transactional processes that a lot of the heavy lifting has been done here, and the new and interesting stuff is going to be turning back to look at the messy human-facing processes that haven’t been touched by BPM yet.

I did drop over for the Savvion panel in the second half of this session, but it was fairly unstructured questions from the audience so difficult to blog although interesting to attend.

My third time today for hearing Janelle Hill speak, but I usually find her to be pretty interesting. This time, her topic is “BPM: A Change from Business as Usual”, taking a look at what’s really new in BPM, how BPM can change the way a company operates, and some BPM use cases.

She started out with a great chart showing what’s new and the implications of each of these points; for example, the fact that processes must be effective and transparent, not just efficient, implies that processes must be explicit and not embedded within applications. In discussing the harmonization of incremental improvement and transformative change, she comes back to the phrase “design for change”, which I’ve heard several times today already; interestingly enough, the subtitle of the Forrester IT Leadership Forum where I’m speaking next week is “Design for People, Build for Change”, indicating that the analysts are really setting the focus on this concept. This is, of course, the heart of business agility: if something isn’t designed and built with the intention that it would be changed frequently, then you’re not going to be changing it much.

Similar to her earlier comments in the opening keynote, the focus is really coming back to the people in the process: the goal of BPM is no longer to automate everything and eliminate people, but also to orchestrate what people are doing. Okay, that’s not really different than the view that those of us dealing with a lot of human-facing processes have had for some time, but it’s a wake-up call for the SOA vendors and IT departments that are focussed on straight-through processing that they have to start looking at the human side.

There’s a lot that has to change with the addition of BPM to a business environment: management based on real-time events, not just transactions; collaboration amongst team members and the team performance rewards that can drive that; and using tools such as simulation and optimization for modeling process improvements.

One important point that she makes about the current state of BPM technology is the explicit process models and their direct link to the executing systems. Although she doesn’t insist on a single shared model, the requirement for real-time round-tripping between the modelling and execution environments is implied in what she says about real-time synchronization between the model and the executing process. In order to make all this work, there are new roles for both business and IT people: the business needs to get involved in some of the analysis and modelling, and IT has to consider how to create reusable components to better enable this.

Hill then moves on to some real-life use cases covering six basic styles of business processes: case management, form-driven/STP workflow, content creation, transactional, guided navigation, and network organization. For each of these, she maps out eight different characteristics of the processes, e.g., what triggers the work to start, and talked about how you can use the combinations of these characteristics to help determine what type of BPMS product that you need. She pointed out that there is no single vendor here at the show (or likely anywhere) that does all six of these types equally well.

She then highlighted the three most common types — case management, STP and guided navigation — and listed common use cases for each of them as well as the expected benefits for each of these patterns. I really liked this section of the talk; it’s a more expansive view than the simpler set of BPM design patterns that I’ve talked about in some of my courses and webinars.

She finished up with a new BPM hype cycle, showing that the technologies are much more mature than the management and methodologies, and some good closing recommendations:

• Start introducing process modelling, analysis and simulation to business leaders to get them ready for what’s coming in BPM.
• Start hiring business process analysts and architects with process skills and experience.
• Look at the characteristics of your process, based on the type of chart that she showed in the presentation, to determine what BPM functionality is required.

The lunchtime address is always a tough one: the speaker has to talk over the sound of clashing cutlery, and half of the audience has to twist around in their chairs to see the speaker and slides, but Gartner tends to keep these short and painless, and features entertaining speakers such as Alan Trefler of Pegasystems.

I was chatting with the others at my lunch table before the talk began, and the person beside me asked my views on some of the vendors. We talked about the pros and cons of product convergence, and I used business rules as an example of something that is sometimes baked right into a BPM product — as with Pega, which is based on a rules engine — but may be more versatile if available as a separate platform. I made a comment along the lines of “just wait, Trefler will tell us why it’s necessary to have rules as part of your BPM”.

Trefler’s focus was on the rhythm of business, the things that can kill that rhythm, and how to fix them:

• Rhythm killer #1: the specification gap, where there’s a huge gap in time and understanding between the business developing requirements, IT developing specifications, then the business signing off on these specifications with the full knowledge that any changes are going to require an equally arduous change order process. The solution: directly capture objectives in the execution environment, that is, use tools that allow the business to create their own models, and have those translate directly to execution.
• Rhythm killer #2: exporting models, which causes the models to be ripped out of the hands of the business and tossed over to IT to import into the execution environment, with no round-tripping. The solution: automate the programming, that is, use a zero-code environment that generates the executable model from that created by the business, either in a shared model environment or in a fully round-trippable environment.
• Rhythm killer #3: no reuse of components or models. The solution: think about enterprise sharing and reuse upfront and design it into the system from the start. Trefler considers the secret of this to be using a declarative rules-based paradigm, of course, proving my earlier comment in his final sentence.

The usual Gartner format is to allocate some sessions for vendors (read: sponsors) and their customers to make short presentations, and I sat in on Sandy Carter talking about IBM’s view of SOA and BPM.

She started with some thoughts on what’s driving business today and the changing business landscape in the global economy, from new technology and business models to new customers and global integration, and how this is driving companies to both innovation and optimization. She talked about some interesting examples of this: McDonald’s is handling their drive-through orders by routing them to operators in the Philippines who interact verbally with the customer to take the order, then key the order into a system that results your burger popping out of the window when you drive up; a perfect example of outsourcing a specific task in a process as opposed to the entire process.

Carter comes from the SOA side and stated up front that BPM must be implemented with SOA; while I certainly agree that SOA makes the implementation of BPM much more efficient, there are plenty of successful BPM implementations that don’t rely on SOA. She showed an enterprise architecture type of view of BPM (although she didn’t call it that), calling out the business view, the process view and the IT view as independent yet interconnected layers.

In looking at their strategy, she covered enhancements to the WebSphere Business Modeler, including improved integration with FileNet and Workplace eForms (by which I assume that she means the FileNet eForms product), which is good news: ever since the acquisition of FileNet by IBM was announced over a year ago, I’ve been wondering how they’re going to properly integrate the FileNet functionality into the WebSphere suite. As an aside, I’ve always felt that the BPM part of the FileNet line should have been moved over to the WebSphere suite from at least a product marketing/product management standpoint, although there would have been technical challenges due to the integrated nature of the FileNet platform: there’s a huge gaping hole where there should be human-facing BPM in the WebSphere line, and FileNet’s BPM would fill that gap.

She also discussed the WebSphere Business Monitor; I’m not sure if this BAM dashboard is the same as the Celequest (now Cognos) BAM capability that’s OEM’d into FileNet’s product line, or if they have a separate BAM offering.

Since IBM is really in the services business these days rather than the software business, she also talked about their new BPM methodology services offering. She showed a video of a very Second Life-ish “interactive management game” in which you run around the game and optimize a business process using BPM and SOA: a “first person shooter” game for business. This is supposed to appeal to the new generation of management that is coming through the universities now; I’d really like to hear the reactions of the students to this and whether they see it as a valid learning mode or just a lame attempt by a large corporation to jump on the Gen-Y bandwagon.