Our friend Jacob contributed a sample of medical rules to the DMN (Decision Model Notation) standardization effort at OMG.  He encouraged me to show how those business rules would look like in our Sparkling Logic SMARTS Decision Management tool.

In this demo, you will see how I created the entire project from scratch, in a business-user-friendly environment, leveraging heavily the use cases he gave me.

RedPen^TM is a patent-pending approach to capturing and updating business rules in the context of data.  Graffiti on the use cases turn into business rules automagically!

Thanks again Jacob for allowing us to demo your project!

We are here on LinkedIn!

You may have joined us from time to time in the Sparkling Logic Community.  Despite its name, the community has been a vendor neutral place for experts in the industry to comment on technology or discipline topics.  We have enjoyed a good run on this social platform but we have decided to relocate…

In the process we have also changed the community name.  Now that Sparkling Logic has a product, it felt more appropriate and less confusing to call it Decision Management Community.  That way experts that work for other vendors and practitioners using other technologies will not feel out-of-place.

The purpose of the community is to discuss best practices and ideas on the usage of the technology.  We have seen in the past great debates between the Chief Architects of the leading BRMS vendors, as well as input from well-known gurus.  I love the genuine and candid input from all, and hope we will continue to have as much passion on LinkedIn.

Please join us!

Yes, you have read the title correctly.  I thought about making predictions for 2012 but my head tends to be far more in the future than I care to admit…  So let me share my vision for Decision Management in 2012, 2013, 2014 and 2015!

Prediction #1: Business Users Rule!

I believe that Business Users will slowly but surely get ownership of their decisioning logic.  As the technologies mature and the business users adopt them more widely, they will rule the decisioning world!

In a very good article on business analytics, Dan Vesset nominated Queen’s I want t all as the theme song for this very phenomenon in the entire industry.  What a perfect fit!

Business users need insight at the speed of thought to assist them with their decision-making.  They need the data, the insight out of the data and the guidance to make the best decision based on that insight.  The traditional model that consist of requesting the reports out of IT or the modeling department is not compatible with their actual needs.  This is where I see the most direct impact on the technology evolution.  Business Intelligence component will make their way into products *and* business users will get used to it routinely.

Getting the insight is not enough though.  I believe that business users will also need to play with their strategies to avoid as much as possible unwanted consequences.  This is where simulation and experimental design will become critical.  By 2015, the number of companies applying champion challenger regularly will probably be on the low side.  This discipline will still be appealing to early adopters and maybe a few in the early majority.

This business empowerment will not happen if we do not change the user interfaces for those products.  If they have to go through IT, they will suffer from the backlog issue we are painfully aware of.  If they need to get an expensive custom-tailored UI built, only the few high-ROI projects will get funding.  Regardless, Darwin’s survival of the fittest theory would select over time the “fittest” which means in our world the “most agile”.  As I like to say, agility is not inconstancy: change is not random.  Competitive Businesses adapt by changing their decisioning to a strategy they have tested as best they could, that they have estimated as best they could, that they trust.  Business users can only do that if they have direct access and control over those strategies.  There is a strong desire to do that of course, this is what has allowed the BRMS industry to grow as much as it has in the past decade.  the barrier to a commoditization of those practices is the steep learning curve.  As an industry, we need to remove this hurdle if we want to allow businesses to embrace Decision Management.

Prediction #2: Analytics for Business

Decision Management aims at bridging the gap between business rules and analytics, combining them together for better decisioning.  The vision is appealing.  That being said, it is hard to institutionalize.  Analytics are owned by the modelers, and business rules by IT or business users in a different group.  Getting alignment between those two groups has been difficult without an executive directive to do so.

In the coming years, I believe that we will see those disciplines getting much more synergistic than they are today.

The enabler is DATA.  Assuming that my Prediction #1 is on target, business users will need data to run their simulations.  Somehow the data problem will get resolved.  If data is available, and the world is definitely becoming more data hungry, big data hungry, then analytically savvy business users will be able to play with it and gain additional insight, possibly developing some predictive analytics.

Let me clarify one thing…  Heavy predictive analytics will not get out of the PhD-packed modeling group.  Those will still work hard on the hard problems.  There are insights though that are not that hard or that change too frequently for the modeling team processes.  This is where business users will have an opportunity to improve their decisioning.  the same way they do not want to go to IT for their routine changes, they will not want to go to modelers for their routine data exploration.

Some applications that work quite well in this paradigm are fraud detection and marketing.  In both cases, trends can change rapidly, faster than you can accumulate the volume of data needed for the deep analytics.  Having other techniques available for the business can allow them to react and either block the new pattern of fraud they see shaping up, or exploit a new fad for their marketing activities.  The idea is to use some “easy” analytics jobs to allow business users to better understand and possibly get a model or some rules to reflect the decisioning they would like to add.

Prediction #3: Back to Decision Support

When we think about decision management, we think about the automated decision services that get embedded into the architecture.  When applications are processed by those automated business rules, it might be approved or declined, but the automation stops with a refer decision.  For some reason, the manual process is always out of the picture.  This is where much has to happen both in terms of technology and in terms of practice.

Decision Support is clearly not a sexy term.  It brings us back a few decades ago…  But I think the term is more encompassing that our dear decision management term.  We probably need someone to come up with a better term now ;-)

The idea is that decision-making is a much bigger activity than just the automated part.  A lot of the assistance that is given in the rich interfaces for business users at rule authoring time, could be given to case workers at decisioning time.  As they are dealing with the exceptions, they would certainly gain by having some simulation and estimation tools to visualize the probable impact of their “instant” decisions.  The impact of a single application may not be big enough you might think…  But, if we could capture the reasoning behind this one decision and apply it to a portfolio, we could better understand the impact of this decision if it was to become automated.  And having it somewhat formalized here would allow a business user committee or process to review the candidate business rules and vet some of them, with or without tweaking, for promotion into the automated system at some point in time.

I think the enabler here will be the new generation.  They are used to social media.  They were born with mobile devices in their hands more or less.  My 4-year-old has mastered a long time ago both our iPhone and iPad, and has figured out how to play the Wii with anonymous web players…  When Gen-Y will enter the workforce, they will push those social habits into the enterprise.  Reaching out publicly in a Twitter or Facebook fashion, likely in the intranet at first, will be second-nature to them.

There is a tremendous value in getting just-in-time information for decision-making…  This fits the “speed of thought” idea that Dan Vesset was referring to earlier.  Whether those knowledge workers will be in authoring mode for a traditional decision management project, or in decision mode in a customer service center, they will leverage better than any of us all the social capabilities that are or will be available in the tools.

Our RETE Series: Part 1, answer and Part 2!

As we are still reflecting on 2011, Carlos and I wanted to continue with the tradition we started last year.
We would like to nominate our

Number One Blog Post for 2011…

And the winner is…

This validates the regain of interest we have all seen for Artificial Intelligence this year — Have you noticed how many people have registered for the Stanford class on AI this year?  It is insane!  More people are joining and want to understand what it is that made this algorithm so prevalent in the business rules industry.  They also want to know more about how it works to better utilize it.  This is a smart move.  It seems that we should keep working on more “how to” posts this year.  I am tempted though to provide more case studies as to how those technologies are delivering value in real life.

Thank you for your readership throughout 2011!

We are looking forward to sharing even more valuable insights in 2010!

It is inevitable…  Business Rules add up over time…  How many rules do you have per project?  hundreds?  Thousands?  We have seen up to 1.2 million of rules in a project!  Maintaining such volumes of rules can be tricky to say the least.

We have been thinking about the different ways you can keep those rulesets from growing like weed…

1. Checking your rules in a different representation can help.  When adding new rules or making modification to existing rules, get the proper context in the way that empowers you the most.  Some have called it “business intelligence on rules”, and it is quite appropriate.  If you can quickly visualize how existing rules function, you can more appropriately fit the new logic without duplication.  You can check out how I leveraged that capability in this video.
2. Have you ever added an “if true” rule in your ruleset because you need a default action?  Those rules cloud the decision logic in the sense that they require the rules writers to know how they function.  If you create this “if true” default action anywhere but the last position in an exclusive ruleset (exclusive means that the rule execution stops at the first rule that is satisfied), then you will shortcut some of your decision logic!  This is the kind of knowledge that is often missed in documentation or internal knowledge transfers, leading to “weird behavior” that might be hard to debug.  How would you do that?  Our recommendation is to use default actions attached to the ruleset.  If no other rule execute then the default action is triggered — you do not have to worry about setting your ruleset to exclusive or some other design you came up with.  It is one less rule to maintain, and it is safe for the generations of rules to come.
3. Do you have “if true” rules that set up variables used in your ruleset?  You can get rid of those by creating variables in your forms of course.  You can also add some pre- or post-processing in the definition of your decision steps or rulesets.  Get rid of those rules that are not real decisioning logic by replacing it with procedural code in there.  The “if true” rules go.  The procedural code is safe from reordering or those pesky priorities that can mess up the original design.  This is a good clean up!
4. Look for the applicability of rules.  Cleaning up business rules start with your business rules but don’t forget to consider the importance of those business rules in the context of your data!  Some decisioning logic may be irrelevant as it never applies in real life…  One person asked the question in our talk at BBC this year “how to retire rules?”…  It starts with knowing those rules are outdated.  Have logs or reports in place to give you this insight.  That is what our customers that keep their rules tight have been doing.

I hope those tips will help you keep up with some of your New Year’s BRMS resolutions!

Caveat: all BRMS products do not support those tips.

Did you guess right on the acquisition rumors that were spreading at BBC 2011 this year?

Corticon announced today its acquisition by Progress software.  One more independent vendors ends up in the heart of a BPM and more platform.  This is a great validation for BRMS technology.  You can’t really fit all decisioning logic into process maps without crowding them.  Another interesting conclusion is that CEP did not suffice either in the BPM platform.  The quotes from the announcement were pretty telling:

Dr. John Bates, chief technology officer, Progress Software said: “Within modern responsive businesses, the need to make informed and accurate decisions ‘in the moment’ is critical. High quality real-time decisions are key to avoid fraudulent transactions, to comply with complex and evolving regulations and to generally make the right decision for the business at the right time. The acquisition of Corticon reinforces Progress’ commitment to deliver operational responsiveness by helping customers build highly agile, responsive business systems with models and tools that maximize simplicity and accelerate time-to-value.”

Dr. Mark Allen, founder and former chief executive officer of Corticon, now a member of the office of Progress’ CTO, added: “[...] rules alone are sometimes not enough; to meet the holistic needs of customers, a number of technology areas need to converge [...]“

Effective Decision Management is all about Agility and Simplicity.  I wholeheartedly agree with that!

Congratulations, Mark!

When you start a business rules project, do you want to spend your time figuring out the right design?  or do you want to focus on the business decision?  Let me illustrate.

When I was at BBC 2011 in Fort Lauderdale, I met a gentleman by the name of Russell.  As we talked about this concept of fluid metaphors aka Visual Logic, he made a brilliant comment:

“users want handrails, not handcuffs!”

What a perfect way to summarize the point I was trying to make in much more words than he did!

Why seeing isn’t everything

In many posts, we have talked about the difficulties of picking one graphical representation for business rules.  Exceptions break the nice and comfortable decision tables.  Decision trees are much better at navigating the decision logic per population segment.  Decision graphs are great for showing the many paths that lead to conclusions.  And textual rules are great to understand the logic one rule at the time.

Being able to turn any ruleset into a decision table, a decision tree or a decision graph is quite helpful for understanding how they all relate to each other…  But what is the point of understanding if you then have to change your view, and start over your thinking process, to find the change you know you need????  Yes there is a point.  No it is not enough!  It would be like staring at the cake from outside the store, through the window…

In this 10-minute video, I demonstrate how you see the “cake” and eat it too:

If you do not have to make the design decision up-front, you can really give more freedom to your subject matter experts and empower them to do what they need to do in the way that they are comfortable with.

No handcuffs.

Do you remember the buzz at Business Rules Forum 2008 when RulesBurst / Haley did not show up?  We all suspected an acquisition and soon got confirmation of the suspicious absence of this “regular” at the show.  They had been acquired by Oracle.

This year again, one of the usual suspects did not come and, since then, the industry has been buzzing quite a bit.  Well, technically the company was there, but the leadership team was not.  We have reasons to believe that our colleagues have been acquired by another platform vendor.  Do not look for the big names though.  It is quite interesting to see the growing presence of BPM / CEP vendors in the Decision Management space, fast acquiring business rules capabilities…

let’s wait and see how long it will take for the rumor to turn into a formal announcement!

On the BRMS side though, we are left with fewer and fewer independent vendors.  This puts more pressure on the vendor locking issue.  The standards are not nearly mature enough to allow for interplay between the rules vendors — which has not been a burning issue up to now.  But as BRMS are becoming an integral part of the entire platform, the initiative of swapping them becomes less of a per-project decision…  and as a result, the current investment in business rules assets might require a port from platform A to platform B, meaning from BRMS A to BRMS B, as companies standardize on those ecosystems…

What are the outlets?  Investing on interoperability standards is the long road.  My personal inclination would be to take a better look at decision modeling environments — but I am biased of course ;-)

Stephen Hendrick, Group VP at IDC, delivers a talk on Decision Management. You may be well acquainted with Stephen, who, among other responsibilities, covers the BRMS and Decision Management spaces for his organization.’

Stephen started his talk with a statement that we’ve been doing event-based decision management for all our life as a species: reaction to events, adaptation to new states, flexibility and evolution being rewarded.  His formal definition is ‘the application of technology to manage the processes that solve the decision problems presented by the intelligent economy” (liberal and approximative quote). The key shift Stephen sees is the transformation of the information economy to the intelligent economy characterized by

• ubiquitous data tracking
• networked interaction
• pervasive analytics
• mobility
• real time interactions
• system to person interactions
• social collaboration and flattening of the enterprise

Of course, this translates (in the way I position it) by the exponential growth of accessible and relevant mostly unstructured business data, and the exponential rate growth of accessible and relevant ever shifting business events. The two trends we discussed with Charles Forgy and Carole-Ann at RulesFest 2011 last week.

This trend both enables better decision management – as Stephen’s states – and challenges most current approaches.

Stephen listed what he considers are the key components of a decision management solution. All typical components are in there, but with a particular emphasis on the social and collaborative aspects to support the management work, as well as on the tracking of business performance to ensure that you don’t just manage decisions as they are, but that you also enable their evolution and optimization.

He made the distinction between

• strategic
• operational
• tactical

decisions, and the different activities involved in the corresponding processes – leading to a categorization of the various types of products and approaches that support them.

Stephen presented an updated IDC Decision Management Model (reference to it here). This model is rooted on work Stephen has been doing over the years. One key challenge Stephen insisted on is the fact that there is no unified Decision Management solution that covers strategic, operational and tactical decisions. Only point solutions exist – and Stephen highlighted how little known some of the tools that address strategic decision management. Many of those have less than 100 customers.

This is an interesting point. In simpler or better known areas such as Business Intelligence similar segmentations do exist. A key question is what the root cause is. It may well be that the nature of the problem is different enough to warrant different solutions – and that the real solution is to have good exchange and traceability between these levels rather than unified approaches.

Stephen went through an additional categorization of the activities involved in decision management

• identity and initiate
• analyze and decide
• act and evaluate
• measure and archive

with a central place to “state change”.

Stephen provided his outlook on the market’s growth potential, which he sees as bright:

• Project-based DM (strategic) – 2009-2014 CAGR 13.2%, at ~$2.7B in 2011
• Tx-based DM (tactical) – 2009-2014 CAGR 17.5%, at  ~$4B in 2011

I am not 100% clear on exactly what products & platforms are included in each section, but combined, we are talking in excess of $6B and a significant growth. It represents around 3% of the overall spend on application and tools. It’s always difficult to figure out the real market share for Decision Management – but the order of magnitude is realistic.

Stephen later went into a catalogue of various decisioning technologies, including rules and various statistical, analytic, optimization and collaborative technologies. One interesting point is that he places Business Rules on the opposite extreme to Adaptive Systems when classified along the lines of decision granularity, uncertainty, complexity and flexibility. Interestingly enough, Carole-Ann and I have a patent (pending) on rules-driven adaptive models – providing a solution to the vexing problem of how to manage the learning process in adaptive analytics.

As a note, part of Stephen’s categorization approach is based on the Cynefin framework – also used by our friends at IBM.

Stephen later went into a categorization of the collaborative decisioning dimensionality, something I had not seen Stephen address before. He made the distinction between various roles: leader, critic, BFF, advisor, friend, etc, and how their various input may be managed – scored, weighed, etc. In his perspective, a lot of the complexity in managing the collaborative process comes from the modality of the decision making (from autocratic to consensus) and the type of decision process (from single-pass to delphi and iterative).

His conclusion captured the key guidance

• if you cannot measure, you cannot manage your decisions
• all decision management activities need to be supported
• feedback-driven decision optimization is key
• the tools to do decision management are mostly here
• making strategic and operational decisions relevant to tactical decisions is complex, and still needs work in terms of achieving consistency and normalization.

To the question on “what will the DM vendors focus on in the next few years”, Stephen’s response was:

• sensor response processing (data and event driven decisions)
• platforms coming together combining technologies addressing multiple aspects of decision management for the intelligent economy

 

Although Healthcare is a perfect fit for Business Rules technology, the providers have been late to adopt the technology.  We have seen many healthcare insurance companies use rules for eligibility or claims processing, but providers are lagging.  I was delighted to attend a session at BBC on Cancer Care.

Michael Katz, BS/MBA at International Myeloma Foundation, gave us a heartbreaking overview of the disease…  incurable but treatable.

The project focused on providing timely information to diagnosed patients healthcare providers.  The solution is available on the internet, with an iPad client available.

Michael shared a demo of the rules they created to detect anemia or bone issues for example.

The iPad application interprets the input data (gender, serotonin level, etc.) by invoking the rules running in a web service.

What’s next?

Nurses have written a 60-page survivor care plan, which they are translating into business rules now.

Similarly, they are translating a 10-step guide for the newly diagnosed (get the correct diagnosis, tests you really need, initial treatment options, supportive care…).