Enterprise Decision Management - a Weblog: May 2006 Archives - BeyeBLOGS.com

« April 2006 | Main | June 2006 »

May 31, 2006

Identity Theft - decisioning in action?

I saw this article on Identity Theft ,and the entirely coincidental press release on Fair Isaac's new Identity Fraud prevention product (Falcon One). Identity Fraud prevention is a classic example of an EDM problem. Why? Well consider the core measures of EDM success - Precision, Consistency, Agility,Speed and Cost. Successful Identity Theft prevention requires:

Precision - don't reject transactions that really are from the person they say they are from, do reject those that are fraudulent
Consistency - identity thieves will use any and all channels to attack so you had better respond consistently across channels
Agility - identity thieves are engaged in an "arms race" with you so you had better be able to keep changing your approach to keep ahead of them
Speed - no-one is willing to wait while you check so you had better be able to check fast
Cost - even though identity fraud is expensive there are bazillions of transactions taking place each day that are real so you can't afford to spend too much checking each one for fraud

Doing this well takes predictive analytics (like Neural Networks) to predict risk accurately, rules to enable agility and policy enforcement, and lots of good data. EDM.

Share:

Posted by James Taylor at 11:13 AM | Comments (0)

May 30, 2006

Book Review: Principles of the Business Rules Approach

This book is one of the classics on business rules from one of the most long-standing authors in the area, Ron Ross. The book is a little more than three years old but, as it is not really focused on technology for managing business rules so much as the general approach, it has aged well.

Ron does a good job explaining what business rules and how to capture them and gives a solid overview of things like fact models, processes and how they relate to rules, and dos and don'ts of rule writing. There is a lot in the book about his particular approach to writing rules which, whether you follow it or not, has some good advice about usage and style for writing declarative business rules.

The book does not go into details on the technology of implementing business rules using a business rules management system but instead focuses on the value of an approach that separates business rules from other kinds of requirements and manages those rules as an asset. A good book to introduce the subject to someone without a technology bent.

There is more on business rules here and some FAQs here and the book can be purchased here.

Share:

Posted by James Taylor at 10:05 AM | Comments (0)

Not another attempt to "fix" requirements!

Regular readers of the blog, and there are getting to be close to 100 of you I think, will have seen my rants on requirements management as a solution for business agility before.

Over on ComputerWorld today I saw another article on companies trying to deliver what their users want by improving requirements management - In Depth: IT Looks to Halt Clashes Between Users, Developers. Now I don't have anything particularly against managing requirements better and I certainly applaud any attempt by anyone to get users and developers on the same page. That said two things need to be made clear:

Business rules are not like other requirements (see these book reviews for more)
Changing business rules are what cause most problems in systems maintenance/evolution (see the whole section on business agility)

Getting better at managing technical requirements like performance and UI design will help. Getting the business and IT to collaborate in the definition of the business rules that drive the system might actually solve the problem.

Share:

Posted by James Taylor at 9:53 AM | Comments (0)

The two kinds of predictive analytics

I saw this article in InformationWeek today - Businesses Mine Data To Predict What Happens Next - and it made me realize that people can be confused as to what predictive analytics means in the context of decision management. In fact it made it clear that there are two uses of the term predictive analytics.

Firstly there is the use of predictive analytics as a kind of analysis done with BI-like tools. This is typically offline and used to inform a knowledge worker.%26nbsp, Essentially this is building a report not on what has happened but on what will probably happen in the future. Many of the examples in the story are like this. Clearly this is an improvement over the usual approach of using reporting and visualization tools to simply understand the past. However, it still assumes a relatively low volume of decisions and that a person with some analytic skills is the best consumer of the prediction.

Second there is the use of predictive analytics as a way to make operational systems "smarter". This is typically both offline - the creation of the models is done offline - and in line - the execution of the models is done during a transaction. The references in the article to how the financial services industry uses predictive analytics mostly refer to this kind of analytic. For instance, fraud detection involves neural networks for predicting the likelihood of fraud. The attraction of this use of predictive analytics is that it applies the "smarts" to every transaction as it happens even if the person involved is not analytically sophisticated (think call center representative) or even if there is no person involved at all (think website or ATM).

When I talk about decision management using predictive analytics and business rules it is largely in this second context. By applying predictive analytics to actual transactions you can get a great return.

Share:

Posted by James Taylor at 9:46 AM | Comments (0) | TrackBack

Interoperable Medical Records - SO WHAT?

In the Federal Times last week there was an article on Your health records online. The government is mandating interoperable medical records by 2014 to try and avoid potentially life-threatening medical errors, such as prescribing drugs that counteract other medicines a patient takes or that a patient may be allergic to and to reduce costs by eliminating errors, duplicative tests and needless hospital admissions.

So will an interoperable medical record do these things. Nope, not on its own.

This is one of those classic cases of mistaking a building block for a solution. It's like saying that a 360 degree view of a customer will result in better customer treatment. It might, it can certainly help but it is not going to deliver any benefit unless it can be used to improve decisions!

Let's take the example used in the article. You are being checked in to a hospital in an emergency and can't answer questions. You have a long, complicated medical history. Having that available would let a doctor identify that you should not be given medicine A but should get medicine B instead. But this is an emergency and the doctor needs to decide quickly. Are they going to scan through your records just in case while you lie there dying? Perhaps, if there are common problems with medicine A. But what if medicine A is fine for almost everyone?

But what if the system they use to prescribe something for you could check the record? Now even the most obscure conflict could be identified. This means using the information to improve the decision being taken - decision management built on a strong data foundation.

What about non-emergency situations. Well even there the doctors I meet are busy and in a hurry so scanning a record is not an option. They either need to be given suggestions based on the record or have their decisions checked against it effectively. Decisioning again.

Don't get me wrong, an integrated medical record can help - it is necessary it is just not sufficient.

For those interested in how to make better healthcare decisions with technology, check out this post on Mark Clare of Parkview Health's approach and a white paper on Using Technology to Seize the Knowledge Management Opportunity in Healthcare by Mark Clare and Mark Pierce at Parkview.

Share:

Posted by James Taylor at 9:27 AM | Comments (0)

May 24, 2006

300 Posts - time for some admin

Well this is the 300th post on this blog - something of a milestone I guess. I thought now might be a good time to remind readers of some useful features of this Typepad blog (and some other ones we have added). First, on the left hand side:

The Subscribe section on the left has buttons to let you subscribe to the feed from this site either as RSS (using Feedburner) or using Email (using FeedBlitz). If you enjoy reading please use this to make sure you don't miss anything
My other blogs are listed below that on the left - check them out for additional content
The categories and archives give you other ways to check out posts. I try and tag entries with the right tags so the categories should be useful

How about the right hand side:

A list of recent posts
Last few entries from the ebizQ blog to make it easy to jump across to that one
Recent comments posted by y'all
Technorati button to show who links here (I'm currently ranked 100,567 for those that track this kind of thing)
Google search, set up to search the blog by default
A blogroll (those blogs I read that seem relevant to the topic at hand)
Some books that I recommend or that others have recommended to me

So, check out the whole blog and send me feedback - jamestaylor AT fairisaac.com

Enjoy.

Share:

Posted by James Taylor at 5:32 PM | Comments (0)

May 23, 2006

Introductory article on business rules

In Follow the rules Marcia Jedd over at AIIM e-doc magazine gives a nice overview of business rules with some best practices outlined.

Share:

Posted by James Taylor at 3:39 PM | Comments (0) | TrackBack

May 22, 2006

Business rules becoming mainstream?

SD Times had an article on how Business rules will play growing role in how companies develop applications and automate processes. Some nice analyst quotes as well as some from yours truly.

Share:

Posted by James Taylor at 4:00 PM | Comments (0) | TrackBack

Beyond BI - Bill gets it (mostly) right...

CIO Insight and eWeek covered a recent memo from Bill Gates on "Beyond BI". As I often talk about EDM being "Beyond BI" I thought I would make a couple of comments on his memo.

"The impact on the workforce is remarkable. Productivity is higher than it's ever been. Buyers can shop the entire world without leaving their desk. Sellers have access to markets that were once beyond reach. The amount of information collected about customers, competitors and markets is unprecedented...."That makes solving information overload/underload a critical task. Fortunately, a new generation of technology innovations is opening the door to solutions that will make it dramatically easier to find relevant information quickly, to use that information to drive intelligent decision-making, "

Well here I completely agree with Bill (I'm sure he'll be delighted to know that) as EDM is all about using technology (business rules, predictive analytics) to drive "intelligent decision-making". I do think that many folks over-estimate the value of making more information available. No matter how easy you make it to consume the information you still assume that the person is able to put it in context and use it. For instance, no matter how clever I get in presenting information about the value of a customer to someone in a call center, I still need to help them put this in context and decide how it should change their interaction with that customer. I call this the "so what" problem:

My call center reps know how profitable each customer is when they call.
So what? Does this change the cross-sell offer, the rules about letting them off charges or what?
My doctor has an electronic medical record of my entire history
So what? Is she going to make a different treatment decision because of it? Will she have the time to read it all or be able to spot the crucial piece?
My insurance agent knows what natural disasters my house it at risk of
So what? Is he going to know how to change my risk-based premium as a result?

Making information more readily available can be important but making better decisions based on it is what pays the bills. Bill goes on to say:

"Resolving the information overload and underload problem will take more than just better search tools. What's required is a comprehensive approach to enterprise information management that spans information creation, collection and use and helps ensure that organizations can unlock the full value of their investments in both information and people"

Well yes, but there must also be ways to turn this better information into better decisions and thus better outcomes. Better informed organizations do not perform better automatically - they perform better if they can make better decisions with that information.

Good start Bill...

Share:

Posted by James Taylor at 3:55 PM | Comments (0)

May 17, 2006

CIO Insight and some thoughts from yours truly

I was "quoted" in a recent CIO Insight column CIO Insight: Techtalk: Fair Isaac Corp.'s James Taylor on automating the decision process. The article set was around new rules for information management and discussed some of the issues around data governance. I think that decision governance is an obvious outgrowth of this trend and that's what I was trying to get at in the column. There's more on the blog about compliance and business intelligence.

Share:

Posted by James Taylor at 10:58 AM | Comments (0)

Congratulations Tommy!

Duke At JavaOne this week, Perrone Robotics' Paul Perrone won a Duke Award in the Emerging Technologies category.

The award described Tommy like this:

Tommy, a project of Perrone Robotics, Inc., is a completely autonomous, Java-technology powered 12-foot-long dune buggy. It was designed for the 2005 DARPA Grand Challenge race in an attempt to be the first robotic vehicle to navigate by itself 150 miles through the Mojave desert in 10 hours or less. , Tommy is driven by Perrone Robotics' patent-pending Mobile Autonomous X-bot (MAX) technology. MAX is a general purpose robotics and automation platform, which dramatically decreases the time and cost involved with developing robotics applications of all shapes and sizes.

Blaze Advisor, Fair Isaac's business rules management system, was used by Perrone Robotics to do offline route planning initially and is going to be tightly integrated into the Max platform going forward. Congratulations to Paul and the rest of the team on the Duke award!

May 15, 2006

Anti-Money-Laundering (AML) and EDM

Aite Group recently published a new report - Anti-Money Laundering Technology: Automating the Haystack Search. Sadly I don't have access to the report but I wanted to take a moment to discuss AML. In the interests of full disclosure I would also say that the report did not include Fair Isaac as we do not offer a specific AML solution. The report expects an increased emphasis on money laundering compliance and notes that AML vendor offerings are improving as emerging needs are identified.

No-one doubts money laundering's potential to impact institutions nor that regulators will continue to look for signs of criminal behavior. The question is how should institutions respond to the problem - by making better use of existing systems or by buying new ones? Personally I think most organizations have the data and systems they need and should be focused not on adding another system but on applying consistent decisioning to their existing systems.

Let's break the AML problem into its two parts - ensuring that one is compliant with the rules (whether or not they catch the crooks) and actually catching crooks/preventing money laundering. On the first point, Eva Weber of the Aite Group said

"Institutions should be asking themselves the same questions that regulators are asking: Am I consistently applying AML policies and procedures? Are those policies and procedures appropriate to the risks my business faces? Can I demonstrate compliance fairly readily?" and
"Regulators are unlikely to penalize institutions for isolated oversights, as long as those institutions have given appropriate thought to their anti-money laundering programs."

So the key issue in compliance is being able to demonstrate that a program has been implemented, that the procedures being followed are appropriate and that the procedures are followed for every transaction. This is clearly a job for a business rules management system - I have written elsewhere on the role of business rules in compliance and this need to demonstrate compliance is a perfect use of business rules and decision automation rather than manual processes or traditional code.

The second problem, actually preventing and catching money laundering, is a little more complex and more akin to a traditional fraud problem. One might use business rules for some of this but one is also likely to build predictive models to enhance them. Neural nets are particularly good at this kind of detection as they learn what is normal and what is not.

To be honest most organizations are letting the government define the rules and then focusing on compliance so this is less of an issue.

If we consider the key AML functionality they list it includes such obviously rule-centric functionality as list checking (something rules can do in batch on interactively) and transaction monitoring (a form of business activity monitoring).

I do not believe there is a single answer when it comes to AML and that it possible to spend a lot of money or a little. I do believe that more organizations should think about applying a decisioning mindset to this problem rather than just buying another application.

I posted in response to an article on AML in insurance recently too.

Share:

Posted by James Taylor at 1:50 PM | Comments (0)

Podcasting!

Well, it had to happen. I am starting to work on regular podcasting and the first fruits of this are now available. There are three to start with:

#1 - An overview of EDM with my friend Ian Turvill

#2 - A discussion of BI and EDM with David Loshlin of B-Eye Network

#3 - An old podcast on SOA and business rules I did for Officer Outlook

To subscribe to the podcast feed, you can click here for Decisions Podcasts.

Enjoy.

Share:

Posted by James Taylor at 12:56 PM | Comments (0) | TrackBack

Explaining analytic models

A posting on the Oracle Data Mining blog made me think about explaining analytics. , Analytics need to be explained because telling a customer (or a regulator) that you took a business decision "because the analytics said to" is not going to fly.

Explainability of a predictive model is essentially the ability for someone to understand the behavior of that predictive model. Often this understanding is in the context of a specified business decision involving real economic consequences. For instance, to understand why a particular applicant was denied credit.

First and foremost, the decision-maker responsible for the decision must sign off on the behavior and performance of that predictive model and must trust that the model is behaving as the expected. In some areas, the model must be provided to regulators in a transparent, mathematically precise form to ensure that it conforms with all applicable regulations regarding that decision area (e.g., the regulation in credit granting that all other things being equal, persons of older age must have a score no lower than persons of younger age). Finally, the model must be fully understandable to the analyst who is creating it.

There are several methods for achieving explainability. Depending on the decision area, the analyst might have to select a model that:

Returns a ranked list of reason codes for adverse decisions to help explain its impact
Permits verification of its behavior by regulators
Ensures that the way in which an output changes does not change direction when an input is changing in only one way
Captures the relevant factors but is still simple enough for non-technical understanding
Has restrictions that ensure that it conforms with relevant regulations
Conforms with the expectations of the analyst and the decision maker

Visualization and other graphical methods can be used to show the results of predictive analytics. The visual representation must clearly show the most important business elements so that those who understand the customers, the business and the regulations can see them.

If you are using predictive scorecards (also known as additive scorecards) explainability can be fairly easy as , a certain set of input conditions creates a score that can be compared to a set of thresholds to identify the decision being recommended

Using a business rules management system to deploy analytics can also be very effective in explainability. Not only can the rules be readily understood by business users, even if the mechanism for deriving them is too mathematically advanced for those users, but as the exact rules fired can be logged for each customer it is possible therefore to look at any given transaction and see exactly how the predictive model played out.

Mathematical models can also be "engineered" so they are robust and respond to changes in the business environment appropriately. One approach is to “weights engineer” when developing models - adjust the contribution of factors to reflect business concerns. For instance a lower weight might be assigned to a characteristic that is not common across customers, and higher weights to those characteristics that are common.

In the end a model has to make intuitive sense as well as statistical sense.

Fair Isaac's Introduction to Predictive Analytics is a great primer on analytics in general, especially the kind useful in operational systems, and my fellow author Rahul Asthana wrote a nice article on "Crossing the analytic chasm" for TDWI. This posting owes much to helpful information from Brendan del Favero in our product management group - thanks Brendan.

Share:

Posted by James Taylor at 11:54 AM | Comments (0)

May 12, 2006

Understanding and Implementing Real-Time Analytics

Michael Gonzalez and I recently gave a presentation on Understanding and Implementing Real-Time Analytics. Michael started off by discussing some trends he sees that lead to a need for real-time analytics:

Drive to operational transformation using insight gained from operations
A focus on business performance management "BI with a purpose"
The move to a real-time enterprise

He then went on to describe some of the issues in a typical BI stack of moving to real-time and how business rules management systems can help in this.

He had a great graphic that showed how there is a loop from descriptive analysis to predictive to prescriptive that enables operational transformation (shown here).

EDM is designed to be a way to deliver this kind of operational transformation by focusing on the business decisions taken by enterprises and on using business rules and analytics to improve them.

Enjoy the webinar.

Share:

Posted by James Taylor at 10:14 AM | Comments (0)

May 10, 2006

Business rules and intelligent user interfaces

I have not blogged much about SmartForms for Blaze Advisor, the extension to the base product that allows for the development of rules-driven user interfaces that take advantage of AJAX and X-Forms so I thought I would do so today. For those of you who hate it when I talk about a Fair Isaac product, please stop reading now!

So, the need for intelligent user interfaces is growing as while modern computer applications have reached new levels of sophistication, web user interfaces have not. In particular the increasingly sophisticated automation of decision-making (aka EDM) is characterized by the need to collect and use large amounts of information. Traditional user interfaces:

Have problems with logic propagation
Cannot adapt to different users
Fail to alter conditions within applications based on context or data

For example the forms complexity in workers comp underwriting involves 4000-8000 class codes, 50 states and unique information requirements for each code.

So what is an Intelligent user interface? Well each component is aware of the semantics of data it captures and rules and constraints applicable to the data apply to the component. Knowledge is imparted declaratively, using business rules and user interaction is modeled on natural adaptive behavioral trends.

Why might you want to do this?

You want to build, interactive smart applications that act as gateway to data capture and decision management
You want to build forms-driven work flows, with business rules controlling form behavior
Build reusable UI components that can be chained across a number of different process flows
You want to streamline an online application process
- Build highly responsive web applications that are data and logic aware
- Reduce the number of incorrectly filled or incomplete from submissions
- Reduce the amount of re-work in complex transactions
You want to use the rules about the data you need to drive an effective “conversation” with your customers
You want to enhance customer satisfaction and deliver an enriched user experience

How can you achieve these kinds of interfaces? Well, combine Business Rules (for agility and consistency) with AJAX (for more responsive applications) and XFORMS, probably one of the best W3C recommendations, for platform and device-independent forms. Fair Isaac calls this SmartForms for Blaze Advisor.

SmartForms extends Blaze Advisor's business rules capabilities by letting you create data-validation-centric business rules and validated web-based forms applications using the same business rules management capabilities as all other business rules in Blaze Advisor. You can build smart interactive decision making applications as well as dynamic “data and logic aware” user interfaces. SmartForms allows multi-step decisioning and interactive questionnaires for Internet-enabled self service.

There are lots of advantages to this - declarative validation rules, integration with back-end decisioning etc, but also in terms of security and scalability. The storing of rules in the browser is a big concern but SmartForms generates XForm pages, XSL transforms, CSS files and the rules in SmartForms are not visible on the client – they are compiled into XPath expressions loaded into memory. SmartForms does not generate any JavaScript code for a specific form (although the SmartForms processor is written in JavaScript). SmartForms is also designed to scale well for numbers of object and numbers of rules per object and the use of standard web server technologies and approaches ensures SmartForms are not the bottleneck.

The end result of all this is that:

Business users create and manage web-based interactive applications
Forms can dynamically change to capture additional data, or change validation rules, depending on previous responses
Forms can be integrated with back-end decisioning services built with Blaze Advisor
Reduced total cost of ownership through improvements in data quality and reductions in maintenance costs
Better data, collected faster, complete the first time

The press release on the new release (6.1) is here.

Share:

Posted by James Taylor at 3:40 PM | Comments (0)

InterACT - A Summary

Well I think we are done blogging from InterACT so I thought I would write a quick summary. There were a ton of good sessions and lots of interesting discussions so the links below represent just a tiny fraction of what went on. My key takeaways were:

Advanced decisioning is not just for financial services any more - insurance, healthcare, telco, CRM, marketing and more are adopting it
The combination of data analysis and human expertise is a powerful one - whether you are combining the two in building new analytic models or by combining analytic models themselves with business rules
Empowering the people who understand the business from a strategic level to impact decisions taken at the operational level is key
There is a tremendous ROI from improving high volume decisions just a little
As Fair Isaac's head of R&,D said, "we are just getting started"

For those of you looking for the links, here they are:

Share:

Posted by James Taylor at 1:27 PM | Comments (0)

May 9, 2006

Live from InterACT (not really): Solving Business Problems at InterACT

This year's InterACT was a showcase for EDM and how rules and analytics work together to make better decisions. As I listened to some of the sessions, however, it occurred to me that InterACT was really a showcase for how EDM, in helping to make better decisions, helps to solve business problems. To reflect this, I came up with the following equation:

S(i) Right Decisionsi = Solution to Business Problems

Let me motivate this equation by asking you to consider the following two scenarios:

Pune, India: A customer calls into a leading credit card issuer's call center in Pune, India and asks that his account be closed down. Asked why he wishes to close his account, the customer responds that he simply has too many cards and needs to trim the number down. The call center employee records this information, but also quickly looks at the customers records and sees that the customer hasn't used the bank's card in months. Maybe the customer is right to close down this card she thinks, but she also wonders, as she processes the customers request to close down the account, why did the customer choose to stop using her company's card? Why not close down another bank card? A thought crosses her mind – what if she could persuade the customer to stay with the company, and close another bank card down? Her thoughts are interrupted by reminders that she is coming up on the allocated time for this customer, and she has 2 more in queue that she needs to get to. She quickly closes the customer's account and moves on to the next call…

12th floor, bank president's office, NY, NY: , The president of the leading credit card issuer is meeting with her top executives and with consultants from a leading strategy consulting firm. The president is troubled that her credit card business is only growing at 4% a year. The industry is growing at an average of 6% a year. This means that someone else is growing faster than the average at her bank's expense. Last night she read the strategy consulting firm's benchmarking report. Nothing in it seemed to indicate that her products were lacking in any way – the benchmarking report instead showed that her bank was making the same loyalty offers, sending out the same number of mailers, and had the same APR's as all the rest. So where was she going wrong? And more important, how could she fix it…?

Interestingly, the answer to the president's problem lies in Pune, India at her bank's call center. It was her idea to offshore her call centers to India – just as all of her counterparts were doing in the industry – to save cost. It never did cross her mind, however, that her call center could be a strategic asset. Or, more correctly, her call center in Pune could be a strategic asset that could be used to solve business problems if it was equipped with EDM and the power to make the right decisions.

Call centers are increasingly the front lines of customer interaction. As such it is at call centers that many of the metrics that drive business performance occur. It is through call centers, for example, that customers can be acquired, for example. It is at call centers that customers who are looking to attrite can be convinced to stay. It is also at call centers that customer satisfaction often occurs – customers often make their decisions on a company based on their interactions with call center representatives. If a business is driven by the sum of performance over all of it's portfolio of customers, it makes sense that call centers will be one of the key interaction points where this performance can be monitored, influenced and improved.

Which leads me back to the connection between the call center employee in Pune, India and the business problem faced by the president of the bank in New York. Certainly, as the example of the customer who had called in to cancel his credit card showed, it is not enough to simply acquire a new customer as a means of growing a business. It is often necessary to do more, particularly in a hyper competitive market like the credit card industry, by encouraging usage of the credit card and by preventing attrition. If a bank is able to do this one customer at a time in the portfolio, the sum of performance of the entire portfolio starts to increase. Once the performance of the portfolio increases, the business problems start to go away. Rapid growth in the bank's portfolio, for example, means that the bank's credit card business will likely grow at higher than the average industry rate.

To grow the performance of the business one customer at a time means that the right decisions have to be made on each customer consistently. The call center employee's instincts were correct when she thought that she should convince the customer not to attrite. If she had succeeded, she would have taken one small step to solving the problem her bank president was facing. The call center employee's problem was that she did not know how, and be able to do so in the limited time she had with the customer.

Here is where equipping the call center with EDM makes it a strategic asset. The promise of EDM is that it delivers the right decisions, derived through the intelligence of Analytics - to the customer touch point through the use of rules. By presenting the call center employee in Pune with all of the right decisions necessary to prevent the customer from attriting, EDM enables the call center employee to meet the strategic goals of the company. The devil, of course, is somewhat in the details. It is possible to personalize the delivery of decisions for each customer through the use of rules. However, it is up to the Analytics in EDM to come up with the right decisions to deliver. The question is – are the Analytics up to delivering this right decision for each customer? For example, if the call center employee wanted to keep the customer from attriting, how exactly should she do it? Should she change the APR? Perhaps offer new loyalty rewards? Or perhaps change the credit line?

At InterACT, two of my colleagues at Fair Isaac, Bill Groves and Barkha Saxena, presented a session on Wallet Share that precisely addressed this issue. Using Analytics, it is now possible to ‘read' each customer who calls into a call center and determine which offer they are most likely to respond to. This means that using Analytics, the call center employee is now equipped to understand which offer – be it a change in APR, new loyalty rewards, or a new credit line – to offer each customer as they call in. These Analytics can be delivered using Blaze Advisor as part of an EDM System.

This paradigm applies to all industries where the performance of the business is the sum of the performance of each portfolio member. If a company is able to make the right decisions consistently on each portfolio member, it can almost certainly guarantee that many of the larger business problems can be solved, and more importantly almost certainly guarantee that it can realize a significant competitive advantage. The key to realizing this paradigm is delivering the intelligence from analytics through rules to ensure that each customer decision is consistently right. In other words, the key to this paradigm is EDM.

Share:

Posted by James Taylor at 8:59 AM | Comments (0)

May 8, 2006

Live from InterACT (not really): When Rules Make the Best Medicine

This excellent session on the use of business rules in improving the delivery of healthcare was given by Mark Clare,VP of Knowledge and Information Management Parkview Health and , Adjunct Professor at Northwestern.

Mark presented on how Parkview Health is using GE Healthcare's Centricity Enterprise (formerly IDX Carecast) product and Blaze Advisor to improve patient care. Parkview is clearly a leader in its adoption of technology to improve care and Mark outlined a list of technology that Parkview is using already.

When it came to their use of Blaze Advisor to manage registration rules, patient care documentation rules, pharmacy rules and other workflow items Mark had some useful observations that seem to me particularly useful for all kinds of rules implementations.

Even simple rules can sometimes help a lot
If you are embedding rules into complex processes, such as patient care, you need to worry about "cognitive fit". That is, the rules should work with the process and the way people think and not against them. For instance, you don't want to stop a doctor with a warning at an awkward moment as they might just click past it.
You need good knowledge engineering to make sure rules are maintainable and execute with the speed you need.
You need knowledge governance to build consensus around rules that cross disciplines or involve hard stops.
You need knowledge value to tie valuable and improved outcomes to your rules.

These four concepts are particularly key in complex environments like medicine but are true to a greater or lesser extent everywhere.

Mark presented his material as part of a Health Data Management webinar also.

Share:

Posted by James Taylor at 5:11 PM | Comments (0)

May 5, 2006

Live from InterACT (not really): The Future of Analytics

This session was given by one of my favorite speakers, Larry Rosenberger. Larry is the head of Analytic R&,D for Fair Isaac and the former CEO. Larry is always fun and often good at challenging the audience to think differently. The prompt for this talks was his recent reading of "Blink" by Malcolm Gladwell and his work thinking about how the kind of almost instant decision-making Gladwell described can be combined with the kind of data/results-driven analytics Larry has worked on for years.

Larry's talk revolved around the challenge to the "better decisions from data" mindset of most analysts represented by the kind of rapid cognition and adaptive unconscious decision-making discussed in Blink. While judgmental decision-making can be fallible (we tend to focus on a single idea too early and seek evidence to support it and have trouble managing randomness or considering lots of variables), the evidence that we are good at some kinds of "snap" judgments in some circumstances is strong. Larry proposed a number of ways to combine the two using Bayesian Statistics as a framework. His examples included:

Using unsupervised learning algorithms to find new outliers and then having experts review them to see which ones are worth tagging as potentially fraudulent for instance.
Use analytics to analyze and instrument subjective overrides of automated decisions to see if some types of override can be built into the automated decision.
Move from a Champion/Challenger approach (where one or two new decision models are compared with the existing one to see which is better) to one where learning strategies are applied to the decision model to maximize the range of potentially new strategies considered.

All interesting stuff and mostly way over my head. What it made me think, though, was that the future of analytics is not in reporting or spreadsheets, visualization or cubes. The future is to find new ways to bring human expertise and data analysis techniques together to make learning mechanisms that start "smart" and get "smarter".

Share:

Posted by James Taylor at 1:10 PM | Comments (0)

Live from InterACT (not really): Wired for Action

Now back from InterACT and trying to catch up on sessions I attended. Michael Chiappetta, Fair Isaac's VP Product Development, gave a presentation on the development of an Enterprise Decision Management architecture for Fair Isaac's application stack. As anyone in retail banking / credit knows, Fair Isaac has a range of market-leading solutions for Marketing, Originations, Customer Management, Collections and Recovery and Fraud detection (credit and identity fraud). Like many companies Fair Isaac built these on a variety of platforms at various times and is now bringing them together onto an Enterprise Decision Management architecture.

Some of the key points Michael made:

Once you have a business rules management solution that supports all your platforms (in his case Blaze Advisor supporting .NET, Java and COBOL) you can create a common repository of business rules across your software products. This improves reuse of core rules, allows for a faster cycle-time for new policies and so on.
Fair Isaac's work to create its own analytic modeling workbench (Model Builder) to allow its 300 analytic modelers/scientists to leverage new and specialized techniques has made it possible to further standardize how models are made and make it easier to inject analytics into all applications.
A common set of decisioning technologies has made it easier to move development resources around and to have experts work consistently across solutions to update rules and design effective decisioning strategies.
Solution connectivity, increasingly of interest to customers, is easier to deliver as fewer distinct technologies are being used.

I enjoyed listening to Michael as it showed again the value of "eating your own dog food" or, as one of my French colleagues likes to say, "drinking your own champagne". When your own company uses your tools and approaches to improve its own solutions, you get some great proof points about it working.

Share: