Lessons Learned from the Real-life Deployment of Decision Management at Scale
8 likes1,375 views
The document outlines lessons learned from the deployment of decision management at scale, emphasizing the importance of decision modeling in managing complex operations. Key takeaways include the accessibility and collaborative nature of decision modeling, the necessity of iterating models, and the idea that decisions should be prioritized over processes. Additionally, it stresses the need to manage decisions effectively to drive improvements and compliance within organizations.
![Pre-Bureau
Risk Category
1 HIGH
2 MEDIUM
3 LOW
4 VERYLOW
5 DECLINE
6 HIGH
7 MEDIUM
8 LOW
TRUE
<100
[100..120[
[120..130]
>130
FALSE
<80
[80..90]
[90..110]
>110
Pre-bureau risk category table
UC
Existing
Customer
Application
Risk Score
© Decision Management Solutions, 2015 14
Process and Decision Models
Process Model Decision Requirements Decision Logic
Identifies decision tasks
Links tasks to Decision Requirements
Models
Specifies Decision Logic or business
rules for automated decisions](https://image.slidesharecdn.com/decisionmodelingdeploymentlessonslearnedbbccon2015-151107231412-lva1-app6891/85/Lessons-Learned-from-the-Real-life-Deployment-of-Decision-Management-at-Scale-14-320.jpg)
![Lesson: Put Decisions In Context
17
©2015
Decision
Managemen
Pre-Bureau
Risk Category
1 HIGH
2 MEDIUM
3 LOW
4 VERYLOW
5 DECLINE
6 HIGH
7 MEDIUM
8 LOW
TRUE
<100
[100..120[
[120..130]
>130
FALSE
<80
[80..90]
[90..110]
>110
Pre-bureau risk category table
UC
Existing
Customer
Application
Risk Score
-amount : Money
-amountLimit : Money
-facilityType : Enum
-id : String
-repaymentBy : Date
-status : Enum
Loan Appliation -address : String
-creditRtng : Enum
-dateOfBirth : Date
-disqualified : Boolean
-disqualificationReason : String
-firstName : String
-id : String
-middleInitials : String
-proofOfIdSeenBy : StaffId
-surname : String
-telephone : TelephoneNo
Person
-annualIncome : Money
-employmentRecord : Enum
-personalDebt : Integer
Financial Profile
-endDate : Date
-salary : Money
-startDate : Date
-terminationReason : Enum
-workType : Enum
Employment
-realisationDate : Date
-type : Enum
-value : Money
AssetLiability
0..* -applicant
1
0..1
-guarantor
1..*
1
1
1
0..*
1
0..*
Process
Data
Decision
Logic
Motivation
Organization
Department
Team
Role
Role
Team
Department
Organization](https://image.slidesharecdn.com/decisionmodelingdeploymentlessonslearnedbbccon2015-151107231412-lva1-app6891/85/Lessons-Learned-from-the-Real-life-Deployment-of-Decision-Management-at-Scale-17-320.jpg)
![© Decision Management Solutions, 2015 18
Lesson: Reinvent Compliance
From documents … To models
Regulation Regulation
Interpretations
Business Requirements
Business Rules
Technical Specification
Code
Technical Rules
Decision
Requirements
Add
Interpretations
Add
Decision LogicPre-Bureau
Risk Category
1 HIGH
2 MEDIUM
3 LOW
4 VERYLOW
5 DECLINE
6 HIGH
7 MEDIUM
8 LOW
TRUE
<100
[100..120[
[120..130]
>130
FALSE
<80
[80..90]
[90..110]
>110
Pre-bureau risk category table
UC
Existing
Customer
Application
Risk Score
Deploy](https://image.slidesharecdn.com/decisionmodelingdeploymentlessonslearnedbbccon2015-151107231412-lva1-app6891/85/Lessons-Learned-from-the-Real-life-Deployment-of-Decision-Management-at-Scale-18-320.jpg)
Lessons Learned from the Real-life Deployment of Decision Management at Scale
- 1. Lessons learned from the real-life deployment of Decision Management at scale JamesTaylor CEO
- 2. Your Presenter - James Taylor CEO of Decision Management Solutions We work with clients to improve their business by identifying and modeling decisions, and applying business rules and analytic technology to automate & improve these decisions. I have spent the last 13 years championing Decision Management. I am one of the submitters of the DMN standard © Decision Management Solutions, 2015 2
- 3. Agenda and Context A discussion of the value of Decision Management and decision modeling to the effective management of large, complex operations - including that of a large, global, financial services organization Agenda Decision Modeling Processes and Decisions Architecture, Governance Agile, Iterative © Decision Management Solutions, 2015 3
- 5. Decision Modeling Some history Business rules and analytics are for decision-making Low-level decision tables are well established Historically higher-level decisions have been implicit The last few years have seen an expansion in up-front modeling of the structure of decisions Decision Model and Notation (DMN) A peer to BPMN, CMMN and in BABOK® 3.0 “… provide a common notation that is readily understandable by all business users” “… creates a standardized bridge for the gap between the business decision design and decision implementation.” ©2015 Decision Management Solutions, OpenRules 5
- 6. An Example Decision Model Break down, decompose, decision-making Show how data is used and where the rules come from Shows dependencies, requirements, not sequence ©2015 Decision Management Solutions, OpenRules 6 Input Data Knowledge Source Decision Decision Structure
- 7. Lesson: Widely Accessible, Collaborative © Decision Management Solutions, 2015 7
- 8. Lesson: Many Use Cases Human Decision-making Documenting human decision-making Improving human decision-making with analytics Training human decision-makers Requirements for automated Decision-making Business rules discovery and analysis Framing predictive analytics Dashboard design Implementing automated Decision-making Completely specifying business rules Acting as a BRMS front-end Orchestrating complex decisioning technology © Decision Management Solutions, 2015 8
- 9. Lesson: Sketchability Drives Creativity Shapes are easy to draw Whiteboard friendly Open to occasional participants © Decision Management Solutions, 2015 9
- 10. Lesson: Networks Not Flow Decisions are often described sequentially The sequence may reflect habit or efficiency Only document logical dependencies © Decision Management Solutions, 2015 10
- 11. Lesson: Find Gaps, Reuse, Duplication © Decision Management Solutions, 2015 11 Decision Requirements Model Reuse is common Use multiple perspectives to find gaps Models are more precise, less forgiving
- 13. Lesson: Model Processes and Decisions Decisions first class object Decisions linked, not buried New process is simplified Independent process & decision changes Decisions are reusable Decisions are explicitly modeled Decisions an afterthought Decisions buried in process Process becomes complex Decisions only evolve with process Hard to share decisions Hard to model decisions precisely Process Only Process and Decision 13© Decision Management Solutions, 2015
- 14. Pre-Bureau Risk Category 1 HIGH 2 MEDIUM 3 LOW 4 VERYLOW 5 DECLINE 6 HIGH 7 MEDIUM 8 LOW TRUE <100 [100..120[ [120..130] >130 FALSE <80 [80..90] [90..110] >110 Pre-bureau risk category table UC Existing Customer Application Risk Score © Decision Management Solutions, 2015 14 Process and Decision Models Process Model Decision Requirements Decision Logic Identifies decision tasks Links tasks to Decision Requirements Models Specifies Decision Logic or business rules for automated decisions
- 15. Lesson: Process Analysis is Not Enough Efficiency Doing the thing right Measures include Time to complete Cost to serve customer Cost to process order Effectiveness Doing the right thing Measures include: Customer profitability Customer retention Fraud losses Efficiency is doing things right; effectiveness is doing the right things. “ ” © Decision Management Solutions, 2015 15 Process Decision
- 17. Lesson: Put Decisions In Context 17 ©2015 Decision Managemen Pre-Bureau Risk Category 1 HIGH 2 MEDIUM 3 LOW 4 VERYLOW 5 DECLINE 6 HIGH 7 MEDIUM 8 LOW TRUE <100 [100..120[ [120..130] >130 FALSE <80 [80..90] [90..110] >110 Pre-bureau risk category table UC Existing Customer Application Risk Score -amount : Money -amountLimit : Money -facilityType : Enum -id : String -repaymentBy : Date -status : Enum Loan Appliation -address : String -creditRtng : Enum -dateOfBirth : Date -disqualified : Boolean -disqualificationReason : String -firstName : String -id : String -middleInitials : String -proofOfIdSeenBy : StaffId -surname : String -telephone : TelephoneNo Person -annualIncome : Money -employmentRecord : Enum -personalDebt : Integer Financial Profile -endDate : Date -salary : Money -startDate : Date -terminationReason : Enum -workType : Enum Employment -realisationDate : Date -type : Enum -value : Money AssetLiability 0..* -applicant 1 0..1 -guarantor 1..* 1 1 1 0..* 1 0..* Process Data Decision Logic Motivation Organization Department Team Role Role Team Department Organization
- 18. © Decision Management Solutions, 2015 18 Lesson: Reinvent Compliance From documents … To models Regulation Regulation Interpretations Business Requirements Business Rules Technical Specification Code Technical Rules Decision Requirements Add Interpretations Add Decision LogicPre-Bureau Risk Category 1 HIGH 2 MEDIUM 3 LOW 4 VERYLOW 5 DECLINE 6 HIGH 7 MEDIUM 8 LOW TRUE <100 [100..120[ [120..130] >130 FALSE <80 [80..90] [90..110] >110 Pre-bureau risk category table UC Existing Customer Application Risk Score Deploy
- 19. Lesson: Decisions Are Better Drivers © Decision Management Solutions, 2015 19 Infrastructure can be a powerful investment Or a bridge to nowhere Decisions are business assets Begin with the decision in mind Drive business rules and analytic technology from decision requirements
- 20. Lesson: Manage Decisions Before Rules Rules only add value once deployed Managing rules requires a structure Not all rules need the same governance, control Managing decisions shows how, where and why to manage and deploy business rules Manage each decision and sub-decision with focused, specific business rules © Decision Management Solutions, 2015 20
- 21. © Decision Management Solutions, 2015 21 Change Rules Management Simplified process models Clear decision models for business users IT-managed Transformation Over-complex process models BRMS Executable Rules BRMS Executable Rules BEFORE AFTER Multiple documents about rules Business Rules duplicated Users must navigate unfamiliar repository One set of Business Rules Users link directly from familiar diagram
- 22. Agile, Iterative
- 23. Lesson: Decisions Are High Change Keep making compliant decisions Track eligibility requirements Competitive discount Change risk assessment Keep the right deal terms 23© Decision Management Solutions, 2015
- 24. Lesson: Model Iteratively © Decision Management Solutions, 2015 24 Build high level models Pick a decision Model it in detail Then document rules, decision logic Repeat for other decisions
- 25. Lesson: Decisions Drive Data Business Objective Decision Decision Requirements Input Data Requirements © Decision Management Solutions, 2015 25
- 26. Questions?
- 27. Takeaways
- 28. Lessons Learned • Widely Accessible, Collaborative • Many Use Cases • Sketchability Drives Creativity • Networks Not Flow • Find Gaps, Reuse, Duplication Decision Modeling • Model Processes and Decisions • Process Analysis is Not Enough Processes and Decisions • Put Decisions In Context • Reinvent Compliance • Decisions Are Better Drivers • Manage Decisions Before Rules Architecture and Governance • Decisions Are High Change • Model Iteratively • Decisions Drive Data Agile, Iterative © Decision Management Solutions, 2015 28
- 29. Thank You JamesTaylor, CEO james@decisionmanagementsolutions.com More on Decision Management at decisionmangementsolutions.com