Neural nets in BPM?

Just saw this article in eWeek about Fuego releasing neural net capabilities in their BPM product.

Neural Network works through a decision activity capability that lets users define a set of variables that can be analyzed for process improvement…Neural Network takes that set of variables and builds a learning activity set that can monitor decisions and suggest behavior to improve the process.

I haven’t heard the term “neural net” much since my days in graduate school when I was slogging through a thesis on pattern recognition; it usually refers to a hardware implementation consisting of a massively-parallel network of simple processors (modelled on the human brain and its highly-connected network of neurons): think grid computing on a very tiny scale. Because these terms are not widely understood, there’s a long history of misuse: in fact, the first company that I worked for after university had the word “perceptron” (a type of neural net) in its name, although we wrote pattern recognition and scientific image analysis software, with nary a neural net in sight.

That being said, I’m assuming that what Fuego is calling “Neural Network” is actually artificial intelligence (AI) or cognitive modelling, although I can understand why the marketing types would avoid the overused “AI”, with its shades of science fiction, and positively run screaming from the overly-geeky “cognitive”. The problem with introducing a functionality that is barely understood in the marketplace (besides having to explain it to your own marketing people) is that the customers have no clue what to do with it, and probably not much time to spend doing the out-of-the-box thinking required to come up with some real business scenarios that have the potential for ROI. If you keep reading the article, you’ll see that the VP of process management at an existing Fuego customer considered “the Neural Network technology” to be “intriguing but not essential”. See the problem? It’s still “technology” in the minds of that customer, not a solution to a business problem.

I think that AI has a great future in BPM, but it’s still very early in the hype cycle. As a natural extension to business activity monitoring (BAM), pushing it into the milieu of semi-automated corporate performance management (CPM), it’s going to be the next “must-do” on BPM vendors’ product plans.

By the way, I wrote this post on my tablet PC (in tablet mode) — the handwriting recognition is really good, although a bit slower than my typing. I would like copy-cut-paste soft keys on the handwriting input panel, however: I had to keep switching from handwriting mode to keyboard mode in order to use Ctrl+C, Ctrl+X and Ctrl+V.

Fractured Language

Yesterday, I was finishing off a presentation for a talk that I’ll be giving next month about corporate performance management, including some of the analytics tools that are used to build things like executive dashboards to display the key performance indicators of a company’s operations as charts and dials. Two tools/metaphors are used a lot: dashboards and scorecards, which both do exactly as they sound. Unfortunately, in my research I found at least one vendor of these products who verbs the nouns, and refers to “dashboarding” and “scorecarding” as the activities of creating these things for a company. Blech.

I felt better after this morning’s daily dose of Savage Chickens.

BPM, BI and performance management

From a Forrester report published last week, “Business Process Management (BPM) Meets BI And Performance Management (xPM) Head On”:

Most business process management (BPM) vendors have integrated their products with business intelligence (BI) for analyzing historical process data; but BPM products should also integrate with BI to support decision-intensive business processes like order management, credit risk management, and sales opportunity management. The problem is that BPM vendors are not currently thinking about BI in this context. They are thinking about automating decision-intensive processes, but current performance management (xPM) products lack the capacity to define and execute business processes. The situation is about to change, however, as BI vendors begin partnering with BPM vendors or even OEMing BPM products. Within the next 12 to 18 months, this trend may even culminate in BI vendors acquiring BPM. As this unfolds, we will be on the brink of a Process to Data (P2D) explosion.

The requirement to integrate BPM data into BI and performance management systems is becoming obvious: I’ve touched on that in several posts in the past, including the integration of BPM with Six Sigma, BPM maturity levels, and the various posts on BPM and compliance that really drive home that BPM is just a part of a much bigger picture. This report talks about the other side of that: having BI and performance management systems launch processes in order to gather information required for decision making.

I’m not sure that I agree with their prediction of BI vendors acquiring BPM, since the ease of integration with many of the top BPM products now would mean that there would be little added value in a BI vendor wedding themselves to a single vendor. BPM in all its forms (including everything down through the integration stack) is a much too pervasive within the average large organization for a BI vendor to assume that they can foist their chosen BPM product on a customer that already likely has at least one other BPM product in use.

Business Performance Management survey

If you caught their webinar on Business Performance Management and Optimization a few weeks ago, you’ll know that ebizQ is putting together a survey on business performance management for their Buyer’s Choice Awards. You can help create and weight the criteria that will be used in the survey if you’re so inclined.

I’m still having a bit of trouble with the whole business performance management thing. The goal is to optimize business performance by aligning what goes on in the business with the corporate KPIs (e.g., customer retention rate). Okay, it makes sense that you focus on doing things that provide measurable improvement to the business; in fact, it’s not only common sense, it’s very similar conceptually to Six Sigma and a number of other business improvement practices that have been around for decades. A somewhat more esoteric goal of business performance management is to allow an organization to make decisions in real time, and become predictive rather than reactive. Lots of potential pitfalls this; it harkens back to the agility problem, which says that organizations aren’t being agile even when they have the capability to do so. Also, BPM definitely plays a major role as an input channel to business performance management, but don’t follow into the trap of optimizing operational processes while inadvertantly blowing your KPIs.

Business performance management is being defined as a combination of methodologies, best practices and technologies, where the technology includes event/process monitoring, management dashboards, rules engines, business intelligence, simulation, collaboration and potentially the kitchen sink. Maybe that’s why we have so many vendors claiming that they’re in business performance management: by this definition, there’s probably fewer software vendors who are not included than who are. In some cases, the technologies of business performance management are collectively referred to as BAM; in other cases, BAM refers only to the event/process monitoring. Personally, I would go for the former definition, but I don’t have a lot invested either way.

My problem is this: what distinguishes the concept of business performance management from that of the decision support systems (DSS) or executive information sytems (EIS) that we built in the 1980’s, except that the technology is a bit more advanced? This is being hailed as the new saviour of business, and I feel a bit like I’m looking at the Emperor’s new clothes.

BPM as part of BAM

A few more notes on today’s ebizQ webinar on BAM. Ms Gold-Bernstein talked about another topic close to my heart, namely that BPM is one of the contributing sources to BAM/performance management, rather than BAM being a part of BPM (as the BPM vendors would have you believe). The term “BAM” was originally coined by Gartner, so they’ve had first dibs at saying what is and is not BAM:

BAM defines the concept of providing real-time access to critical business performance indicators, along with the supporting information to improve the speed and effectiveness of business operations.BAM is accomplished by monitoring multiple systems, creating real-time dashboards, and using context and rules to detect the occurrence of a pre-defined set of circumstances.

They list BAM technologies as including BPM, integration middleware (arguably part of BPM under Gartner’s own definition), BI, dashboards with KPIs (which I would consider part of BI), and IT operational management (ditto). Since BAM is defined as a concept and is linked to all of these technologies, there are a lot of vendors from all different areas scrambling to get into BAM magic quadrant — not unlike what’s happening with BPM vendors ever since Gartner lumped together all process-related technologies as “BPM”.

To confuse things further, Gartner’s report on the convergence of BPM and BAM lists three main areas of overlap, and therefore potential conflict:

  • BPM acting as “BPM+BAM”
  • BPM serving as BAM’s response mechanism or recipient
  • BPM ? or business process analysis (BPA) ? serving as a passive analytic/visualization model for BAM

Prior to Gartner defining BAM, there was performance management, which is more focussed on the BI side of the equation, including technologies such as BI, dashboards and, lately, CEP (complex event processing). Although the goals of performance management are fundamentally the same as BAM (business alignment, real-time KPIs), the scope is narrower by excluding BPM and middleware technologies.

Somehow, the concept of performance management as pure business intelligence makes more sense to me than including (rather arbitrarily) some of the technologies that produce the data that feed into the performance management. If BPM is included as one of BAM’s technologies, why not databases, or CRM, or any other technology within an enterprise that produces data that may be of interest to management? In fact, if there’s a technology within an enterprise that doesn’t contribute data to performance management KPIs, why is it there?

BPM and measuring performance

A week ago today, I was in a tutorial by Roger Burlton of the Process Renewal Group on Enterprise Business Architecture: Strategy, Process and Capabilty Alignment while at the BPMG conference in London. He made a great analogy regarding performance measures: knowing the final score doesn’t tell you if it was a good game. (Although after once watching the Blue Jays lose 22-2, I posit that you could have told that it was a crappy game by the score alone.) Roger was one of many people at the conference who spoke about BPM in the context of measuring business goals. To quote Terry Schurter, who I heard speak the following day at his session BPMS – Selection by Business Value, “goals that can’t be measured aren’t goals”.

All this shows that BPM is finally creeping out of departments and into the mainstream of the enterprise. I read a post about a CEO’s view of BPM as discussed at a recent CEO roundtable today on the Milestone Group’s Thoughts On The Tech Industry blog that really nailed it:

BPM, or Business Process Management is the fasting growing segment in the BI / Data Management market sector. Growing out of departmental process management initiatives, the key driver now is to integrate across the enterprise, and to provide scorecards, reports, and other “C Level” deliverables with more confidence, so that predictive modeling and optimization initiatives are really based on the most complete and highest quality set of data available in the enterprise.

The CEOs consider BPM as a feed for enterprise BI/performance management, which is completely accurate — after all, why else would you be doing BPM if not to improve performance, and why would you be doing anything to improve performance if you weren’t also including it in your enterprise performance measures?