A Taxonomy for Case Management – 3

Building the Taxonomy

By combining the characteristics in different ways we can begin to build thef taxonomy for case-based applications.  Here we will give three characteristic examples, which we can think of as three different case species.

Example 1 – Travel Request

Case Species: Structured/Deterministic/Individual

Defining Characteristic Values Comments
Trigger Manual The traveller fills out the request on the web
Information Structured No documents needed to request travel
Processes Deterministic With the exception that the manager may use judgment in approving/rejecting certain trips.
Perform Mode Individual No collaboration involved
Close Archive No need to restart

Example 2 – Auto Insurance Claims

Case Species: Unstructured/Non-Deterministic/Individual

Defining Characteristic Values Comments
Trigger Manual Policyholder places call to agent, who starts case manually.
Information Unstructured Insurance policy document, accident photographs, body-shop estimates
Processes Non-Deterministic Claims adjuster needs to make decisions based on the nature of the accident
Perform Mode Individual No collaboration involved
Close Many need to restart For example, if the repairs turn out to be defective

Example 3: – Grants Management

Case Species: Unstructured/Deterministic/Collaborative

Defining Characteristic Values Comments
Trigger Automatic Grants are received via HTTP
Information Structured and Unstructured Grant proposal is the key document
Processes Deterministic Grants manager can decide to send the grant proposal to multiple reviewers and can approve at any time.
Perform Mode Collaborative Often collaboration between grant reviewers is important
Close Archive Once the grant has been approved, there is no need to restart.

Using the Taxonomy

This is all interesting but what good does this taxonomy do for us? Why bother? There are three reasons to think about case-based applications in this way:

  • To create a basis for discussing requirements
  • To realize what technologies will be most relevant
  • To provide a starting point for change management

Using the Taxonomy to Discuss Requirements

Requirements are hard. They remind me of the blind man and the elephant: everyone sees a different aspect of the solution and they are all right – and all wrong. The problem is to see the whole solution and not just a part. This is where the classification fits in. It provides a balance between one-sided perspectives. And it is a good way to think at a high level. For example, you know that the process is deterministic but that collaboration will be required. It enables you to get general agreement on the shape and structure before you have had a chance to dig into the details.

Using the Taxonomy to choose Technologies

You know from the taxonomy what technologies you need to explore. Some solutions will be process intensive while others will be data intensive. Some solutions will require a means of collaboration, while others won’t. Some solutions will need to be archived and reopened. These are important things to know early in your development cycle, even before you have detailed requirements.

Using the Taxonomy for Change Management

One of the great things about a new solution is that it changes the way people work. If you do it right, it will promote greater efficiency, productivity, and cooperation. But technology does not achieve these goals on its own. Critically important is a well orchestrated plan for education, training, management focus, and measurement. The taxonomy can help you spot which elements of change management will be the most important and the most challenging. This gives you time to prepare the change management process while development is in progress. All in all, knowing what kind of creature you want to create will help you think through the project, explain it to management, and get people on board early in the development cycle.

Aside | Posted on January 6, 2012 by | Leave a comment

A Taxonomy for Case Management – 2

Posted on December 1, 2011 by

Information

The information that underlies a case typically consists of structured data and various kinds of supporting content – documents, photographs, images, and so forth.  Some cases are data intensive, having few or no documents. Other cases are document centric. For example, at our company we use a Travel Application based on xCP that includes no documents, just data. On the other hand, a marketing application might very well be focused on writing, approving and publishing sales collateral. Of course the management of unstructured content is based on the notion of metadata – so we always have some level of structured data.

Process

The information artifacts undergo changes as the case progresses. Information is consumed, entered, reviewed, and changed. Documents get written, signed, and approved. Actions are taken, like rejecting a request for summary dismissal in a hearing. These constitute the process dimension, the aspect that exists in time. The processes consist of elementary units called activities, which can be performed by people or systems. We can classify the activities as deterministic or non-deterministic. All that means is whether the activity is planned in advance at design time, or is decided dynamically, at run time. For example, a case worker may make a decision on the spot that could not be known in advance.

Needless to say, many cases require a combination of deterministic and non-deterministic activities. For example, human decision making is often hard to reduce to a simple set of rules and requires flexibility and judgment, but we want our utility processes (those processes that are performed in the background by automated systems) to be completely predictable and deterministic.

Performance Mode

Another process-related issue relates to the modes in which people perform case activities. In some cases each performer carries out tasks as a single individual. For example, my supervisor approves my travel request. However, in some other situations it may require two or more people to work together. Perhaps two designers are collaborating on the data structure of an application. Or perhaps a committee needs to approve a proposal.

Close

Finally, we need to consider the ending of the case. In almost all situations, it will require a decision by an authorized entity. For example, in a criminal courts case, there is a verdict – innocent or guilty. Your travel request is approved. The main difference is what happens after a case is closed. In some types of cases, the case is destroyed. In others the case is archived. Some types of cases can be restarted in the future; others cannot. Of course, if it is a requirement that the case can be restarted, then that will have implications for how we store the archived information, for security, and for retention policies.

Summary of Case Characteristics

 

The following diagram summarizes the discussion so far:

Posted in Case Management | Tagged , , , , | Leave a comment

A Taxonomy for Case Management – 1

Posted on November 4, 2011 by

Biologists use taxonomies to make sense of the biosphere. It’s the old story of “divide and conquer.” By breaking complex things down to simpler ones, we can understand relationships that might not otherwise be so obvious. Like a biologist, we have to look for common defining characteristics among the creatures we want to study and classify.

Taxonomy is often deprecated as “mere description or subjective.” Even biologists sometimes have to defend the importance of taxonomy. Consider the following quote:

“Given the fundamental necessity of reliable taxonomy for biology, it is surprising that taxonomy is so frequently misunderstood. Even former disciplines of “natural history” with their roots in the taxonomic tradition frequently attack taxonomy on false grounds in spite of the dependence of such fields upon credible taxonomic information.”

What makes a good Taxonomy?

Here is another quote from the same article:
“A good species description is a standard in perpetuity. High quality descriptive taxonomy publications are cited and studied and referenced for centuries.”

Following is a representation of a biological taxonomy, at the highest level – the kingdoms.

Notice how simple it is.

According to the Green Chameleon,

“The great military writer von Clausewitz speaks of the importance of the “coup d’oeuil” for the experienced general. By this he means that with one “cast of the eye” over the military situation, the general can immediately grasp its implications and start to anticipate appropriate courses of action.”

A good taxonomy has to be:

  • Comprehensible – easy to understand, typically this means hierarchically organized
  • Comprehensive – cover all the species under consideration
  • Explanatory–provide insights that can have diagnostic and explanatory value

Definition of Case Management

To start off, let’s get agreement on the creatures we are considering. What exactly do we mean by case management – or, better yet, “case-based applications?” A case exists in space and time. By space I really mean the matter of the case, which can be documents, data, software, or even physical objects. (Think of a glove used as evidence in court). The spatial dimension encompasses all the objects with which the case will deal.

In this context the notion of time refers to the dynamics or progression of the case. Every case has a starting point. The matter of the case unfolds over time, and undergoes transformation, ultimately arriving at an outcome, conclusion, decision, or action that completes the work of the case.

Case Characteristics

For our purposes, cased-based applications will be characterized in terms of the following four common elements:

  • Trigger
  • Information artifacts
  • Orchestration
  • Case Closure

We’ll expand on these four elements in our next posting

Posted in Case Management | Tagged , , , , | Leave a comment

Why Big Data Matters to BPM (and especially to Case Management) – Part 4

Posted on October 26, 2011 by

Let’s  look at the case lifecycle in more detail – to understand exactly how analytics impinge upon the operational case. Recall the diagram:

1. Know When to Act

What triggers a case? Systems watch data and send signals when significant events occur. For example, a Holter monitor indicates that the patient is experiencing arrhythmic heartbeats. A stock tracking system sends out an alert that the Dow Jones average has dropped below 10,000. An inventory management system signals that the number of iPads in stock has become dangerously low.

Knowing how to respond to such events is precisely the domain of case management. Think of tax audits, mergers & acquisitions, anti-terrorism, healthcare, and customer churn. In some cases you are looking for something subtle, like patterns of events, such as a recurring dip in the price of gold on Tuesday mornings. Such advanced forms of monitoring are the domain of Complex Event Processing, which performs event monitoring, reporting, recording and filtering.

2. Inline Analytics

A case includes some form of decision and often many decisions. Should I approve this loan application? What raise should I give employee John Kowalski? Some decisions are based on subjective judgment but increasingly important decisions need to be based on hard data. This is where big data and analytics play a critical role.

What is the meaning of “inline analytics?” It is simply a collection of reports, graphs and other analytical capabilities that help the case worker to understand the data and make the best possible decisions on the case in progress. There are different levels of analytics as shown below.


Here are a few examples to think about.

Calculation

  • Calculate insurance premium
  • Compute total cost of products customer has ordered this year

Visualization

  • Graph, chart, gauge
  • Scatter plot
  • Heat Map
  • Geographic maps.

Here’s a colorful visualization from Time magazine:

Classification

  • Case type is simple, average, or complex
  • Customer is at risk for churn
  • Customer is entitled to extended warranty

Estimation

  • The income of the customer is greater than $200K
  • Mean time to failure for this product is greater than 1 year
  • It will take 4 days to ship the product the customer ordered
  • The likelihood is 75% that this claim is fraudulent

Recommendation

  • Target promotions based on purchase history
  • Recommend that the customer perform preventive maintenance
  • Reject mortgage application
  • Recommend diet and exercise program to patient

Prediction

  • This customer, J. Bananas, is likely to default on his loan application
  • Amount of rain for September will exceed 5 inches
  • Order volume will be down 10% in Q4
  • The S&P 500 will reach 13,000 in November

The first two types are deterministic whereas the others result in probabilities. As you progress the level of analytical sophistication increases.

3. Business Rules

With analytics you can create “fuzzy business rules” for example, the likelihood is 80% that customer will not repay the loan, so it may be the right decision to reject the loan application. Rules can be exercised through human judgment calls by a case worker or they can fire automatically.

4. Learning from Experience

Inline analytics depend on background analytics. A background analytical agent discovers unknown patterns. For example, to make a recommendation requires that a data mining algorithm detects relationships between variables. (The classic example is a market basket of beer and diapers).

So the exhaust from case processing creates valuable data that needs to be stored and analyzed. Much can be learned from experience. Here are three examples.

  • Operational Metrics indicate how well your business is performing, whether you are meeting your goals. It gives you hard data on Service Level Agreements and Key Performance Indicators, without which you can’t manage the business.
  • Predictive Analysis enables you to discover which inputs contribute to successful outcomes and which result in failure. For example, you can predict which customers will default on their mortgages or determine which students are on the path to failure.
  • Process Improvement means that you should change the way you handle cases, either by restructuring the processes, re-training employees, or adding automation.

Learning from experience means that you can continually change and improve the algorithms that you use to support decision making based on real metrics.

Summary

The explosion of big data is the outcome of recent technology innovations. Increasingly, businesses will need to incorporate mobile data, social networks, digital video, and sensor data into their case management processes. They will need to extract semantic content from documents and use it to make decisions. The big data lifecycle makes it easy to recognize the touch points between big data and case management. Among these are signals to act, better decision making, and learning from previous cases. By applying big data and analytics, businesses can improve the timeliness, efficiency, and quality of their case management.

Posted in Case Management, Data and Analytics | 1 Comment

Steve Jobs: An Appreciation

Posted on October 7, 2011 by

Steven Paul Jobs is dead at 56. I will not call it the end of an era. Jobs introduced a new spirit of innovation that I hope will live long after him. His greatest contribution was not to develop a breakthrough product or to lead a company to market domination but to introduce a profoundly innovative spirit that has permeated a wealth of products – at Apple, Pixar, and other companies.

It’s  not easy to characterize his spirit. I think it is reflected in three main characteristics:

  • Dare to be Revolutionary
  • Create Joy for the User
  • Insist on Beauty

My first computer was an Apple. It charmed and delighted me. It is hard to remember how daring the first Macintosh was – how completely unlike any computer that preceded it. The windows interface, the mouse, and the graphical capabilities were innovative and unprecedented. Even more revolutionary was the iPhone. These were not the products of focus groups. They did not originate with computer scientists and programmers, but were closer to magic, inspired by artists, poets, historians, and other designers outside the engineering world. As Steve Wozniak put it, “Somehow he had the ability to think out new ways of doing things, not just improve what we already had, but to do them in a totally different way.”

In the 1980’s Mac users constituted a cult, and for good reason. I remember how shockingly easy it was to use AppleTalk compared to the computer networks of other vendors.  Apple products continue to engender appreciation and loyalty. There is an intense level of customer loyalty that reflects the joy of their users. iPad users certainly feel this today.

Steve Jobs was an esthetic perfectionist. He was personally involved every aspect of product design. He would reject a design because it used the wrong font or had the wrong kind of screw. Every time I use my iPOD, I marvel at how sleek it is, how perfectly formed it is. Brancusi would approve.

This innovative technological spirit grew out of Steve’s life experiences. He started life as a child adopted by working class parents. He dropped out of Reed College. Then  Steve Wozniak and Steve Jobs created Apple computer in their proverbial garage. He lived through, and learned from, the social revolutions of the sixties and seventies. He grew from his failures and was enriched by his spiritual quests. All this gave him the courage to insist on the insanely great. It changed the world. May his spirit never be lost.

Posted in Uncategorized | Tagged , , | Leave a comment

Why Big Data Matters to BPM (and especially to Case Management) – Part 3

Posted on October 4, 2011 by

So far, we’ve looked at Big Data as a general phenomenon: where it comes from and what makes it interesting. Now let’s tie it in with Case Management.

The Big-Data-Case Lifecycle

One helpful way to think about the use of big data is to view it within a lifecycle, as shown below.

Three Levels of Lifecycle

This lifecycle is very generic. It can be applied at three different levels of business decision: strategic, tactical, and operational. In terms of prevalence, strategic decisions are the rarest and operational decisions are the most prevalent as shown below.

A simple (perhaps oversimplified) way to think about the three levels is that the operational level carries out business processes, the tactical level designs and changes the processes, and the strategic level decides which processes to perform. Our focus here is on the operational level rather than the strategic or tactical level.

The Operational Level: The Case

The operational level is where things happen. A case starts – you acquire data, you analyze the data to understand what it means. With the insights that come from this inline analysis, you make decisions in your case. Should I approve this loan application? Or not?

These decisions themselves are based on implicit or explicit business rules and may be provided by a Business Rules Engine, human judgment, or a combination of these. Decisions are translated into action through BPM, which is effectively the “action engine.” As we shall see, the results of these actions are themselves consumed as data and ultimately become fodder for further analysis, enriching the analytics and closing the loop.

Bottom line: analytics are what make a case intelligent. They help the case worker make better decisions and reduce the risk of failure.

In our next posting, we explore the lifecycle in more detail.

Posted in Case Management, Data and Analytics | Tagged , , , , , | 2 Comments

Why Big Data Matters to BPM (and especially to Case Management) – Part 2

Posted on September 26, 2011 by

From our last post we’ve got a pretty good idea where all this Big Data is coming from. In this post I argue that size matters. Then I turn around and argue that size doesn’t matter. As Emerson said, “A foolish consistency is the hobgoblin of little minds.”

Point: Size Matters

As you move up in scale from megabytes to petabytes to exabytes, the world changes. Suppose you are a product manager working for a large pharmaceutical company. You are developing and marketing a new drug, which you hope will be highly profitable. You detail the costs and benefits and write your business case. You need to understand the competing drugs, as well as the risks with your new drug. It takes months of research.

Now, imagine that you had access to big data. Your big data repository encompasses millions of items on competing drugs, together with details on costs, patient needs, adverse effects, and drug interactions. You can tap into intelligence on the moves and strategies of your competitors. The data speaks to you through the analytics. You do a large-scale regression analysis to find out the potential demand in five years. Your business case becomes radically different and more convincing. The amount of data has changed the game, allowing you to delve deeply into more dimensions and to plan with greater rigor and certainty.

The same story applies to other industries, such as mortgage lending, public sector, claims processing, education, and tax auditing. Having a critical mass of information transforms what you can do. Size matters.

The New Face of Data

Big Data is multidimensional. Data volume is not the only dimension, and not always the most important one. Here are five dimensions in which big data is clearly not your father’s Business Intelligence.

1. Reach – Traditional business transactions encompass a single operational system. Big Data often reaches out into sources across the world. For example, you may want to tap into data from thousands of Twitter feeds or from millions of mobile phones. The classic example of wide reach is searching the web a la Google or Yahoo. (According to one estimate,  Google uses 900,000 servers to handle this data).

2. Granularity – With big data you can have unlimited drilldown, so a business can inspect every click on a web page, an economist can track minute fluctuations in the macro-economic environment, or an energy scientist can analyze seismic and geophysical data. Some decisions require this level of fine-grained detail.

3. Variety – Traditional BI looks at structured data, typically from a relational database or data warehouse. Big data extends the variety of data to include new types of information, like text, video, and social media. Bringing these new information types into the picture adds contextual understanding.

4. Velocity – Traditional databases cannot deal with data that is constantly changing. New techniques are required to handle high-velocity data. In real-time database systems data arrives in real time and is analyzed quickly. Classic examples are stream databases that respond to the ebb and flow of the stock market or systems that monitor, analyze, and respond to a patient’s vital signs as they occur.

5. Fluidity – Unlike traditional BI, big data is not always fixed in structure; its content can be complex and constantly evolving. New innovative database techniques, like NoSQL and key-value stores, go beyond static data warehouses to enable dynamically extensible schemas. In many of these new architectures, the traditional requirement for consistency is relaxed from “always consistent” to “eventually consistent.”

We are living in an era of great innovation in data management. In fact, Merrill Lynch calls this the “Golden Era of Database.”

Counterpoint: Size Doesn’t Matter

The big data revolution has changed our way of thinking about business assets. In days of yesteryear, business assets were as solid as bricks and mortar. Think US Steel. With the emergence of the Internet, assets became virtual. Advertising dollars follow eyeballs, which follow web content. Think Google. Today all forms data – not just customer data – have become strategic assets. In fact the value proposition of some companies is based entirely on acquiring and selling data. For example, Apple lets you access CD jacket information when you’re ripping a CD. Gracenote is the company that makes this possible. The data is pulled into iTunes from the Gracenote database. It includes track lengths, track titles, artists, and various other album details. Customers can add their own CD information using a simple form.

Now this adds new data that is then available for the next customer. And the Gracenote database grows and grows.

What’s the point? We’re talking about a new approach to data as a business asset with new and innovative ways of extracting value from it. Who cares if if it’s “Big Data” or “Medium Data” or “Small Data.” In fact, one important technique is to start by reducing Big Data problems to Small Data problems. Come to think of it, maybe we should stop calling it “Big Data” and just call it “New Data” (hmm, but that name would itself soon become passé).

We haven’t talked about case-based applications yet. That’s the topic for our next post.

Posted in Case Management, Data and Analytics | Tagged , , , | 2 Comments