A mind-map of Decision Intelligence

Let’s understand the concept of rational decisions. The use of decision intelligence systematically and at scale converts this skill into an organizational asset.

1. Different kinds of decisions
   1. Organizational decisions: there are shared values (e.g., maximize profit) so the objectives are clear and often measurable, guiding the decision-making and learning processes
   2. Individual decisions: values are individual-specific and vary with time and person, dispersion in outcomes kills time (can be socially costly to differ from a cohort), possible to make decision support models like “in general if you do x the effects are y”. Used for determining medical treatment, buying life insurance, managing investments of time and money, career choices, performance coaching, etc.
   3. Political decisions: where values and objectives are up for debate, e.g., in democratic governance decisions.
2. Decision Needs and Decision Layers. There are four kinds of decisions: strategy, capacity, scheduling, and execution.
   1. Strategy Decisions. Navigate the course of the organization in the context of the industry and set the organization’s objectives.
   2. Capacity or Capability/Size Decisions. Capabilities perform functions such as sales, service, manufacturing, etc. Decisions are to reduce, increase, add, remove, outsource, etc.
   3. Scheduling & Allocation. Decisions that set the schedule and assignment of workloads, such as job-scheduling optimization or driver/trip allocation.
   4. Intelligent Execution. Decisions taken by the people in the flow of their assigned work.
3. Decision Chains and Dependencies. Decisions are interconnected, and it is useful to trace and model the main interconnections to assure alignment.
   1. Cascading decisions, cascading down (from strategy to capacity to scheduling to workflow) or cascading back up (from workflow to scheduling, etc.)
   2. Sequential decisions (such as assigning truck drivers to loads) where the decisions occur repeatedly and what’s optimal in one cycle can be suboptimal over repeated cycles. Ref the Castle Lab in Princeton University.
4. Evaluate & Evolve. Analyze results and update the playbook.
   1. A/B Testing, Design of Experiments (DOE), and Natural Experiments
   2. PDCA (Plan, Do, Check, Act), DMAIC (define, measure, analyze, improve, control), and Continuous Improvement (Kaizen)
   3. OODA (observe–orient–decide–act) by Colonel John Boyd
   4. Adaptive Enterprise by Stephan H. Haeckel
   5. Decision Cycle by Rahul Saxena. Includes “playbooks” that go from decision-opportunity to advice to execution, nested in “evolvers” that evaluate and evolve the playbooks.

There are three major types of Decision Intelligence Systems: (1) data supply chain: data-pipelines, data warehouses, report-makers & dashboard-makers; (2) intelligent behavior: forecasting, optimization, simulation, decision modeling, workflow managers, rules-bases, AI agents, etc.; and (3) user experience for intelligent systems, currently exemplified by LLM-powered chat that can access data and provide analyses, summaries, etc., which will evolve to transform the user interface, interactions, workflows, conversations, and the overall user experience.

1. Data Supply Chain
   1. Data Pipelines: fetch, transform, and store. Extract-Transform-Load (ETL) and Extract-Load-Transform (ELT). Source data from OLTP systems (ERP, CRM, eCommerce, etc.), SCADA, PLCs, RFID, GPS, sensors, chat, social media, survey, votes, documents, images, video, audio, benchmarks (including stock-market data and other public metrics), and analytical outputs that serve as inputs for further analyses.
   2. Enterprise Memory composed of structured and unstructured data stored in data-warehouses, data-marts, data-lakes, and file systems, etc.
   3. Data Quality, Metadata Management, Data Tie-out, and Data Stewardship
   4. Dashboards, Reports, and Signals
2. Intelligent Behavior using Playbooks, Situation Awareness, and Conversational Intelligence
   1. Playbooks for Intelligent Decisions
      1. Playbook Creation, Assessment, and Updates
      2. Playbook Operations & Adoption: Analysis to Advice, Advice to Decision, Decision to Execution, Execution to Results, Evaluation to Evolution
      3. Playbook Inventory and Introspection. Track which decisions have playbooks, and how the playbooks align. This is needed to close gaps in the playbooks, to find places where playbooks don’t exist, and where playbooks are misaligned between different workflows.
   2. Situation Awareness to monitor the organization’s ecosystem, locate hotspots (opportunities and problems), alert the commanders, and help direct the response.
      1. Perception: monitor, detect, and report events, system states, locations, and conditions. Requires the Data Supply Chain to collect and process these elements.
      2. Comprehension: integrate and synthesize perceived elements in the organization’s context. To understand the meaning and significance of the situational hotspots in relation to one’s goals, you need to combine, interpret, and retain information to form a holistic, coherent picture of the situation. Requires the Data Supply Chain to convert the incoming elements into a coherent semantic model to drive comprehension and generate signals.
      3. Simulation: project the signals into scenarios to anticipate future events and dynamics. Enable timely decision-making.
   3. Conversational Intelligence to hold specific and accurate data-driven conversations. Research shows that intelligent conversations lead to intelligent thinking. The system can  hold an intelligent conversation: provide information, spark insights, and enable intelligent behaviors by its conversational style with precise data-driven and context-aligned responses.
3. Interaction Intelligence that provides user-friendliness and personalized assistance. This functionality makes the system pleasant-to-use because it is aware of what’s needed by each person at each interaction.
   1. User-Aware: takes inputs from the organization structure, decision rights, roles, user-profiles, and user-preferences to provide the right user experiences.
   2. Data-Aware: accesses the Enterprise Memory to know the current situation, trends, history, and context. Provides relevant, reliable, and accurate assistance.
   3. AI-Composed: to compose, customize, and update the user experience to make each session maximally satisfying for the user.

What are the components of decision intelligence expertise?

1. Problem Framing & Definition. Modeling objectives, constraints, choices, structure, & relationships. Determining data requirements.
2. Data transformation to meet the data requirements of the problem at hand. Management of data quality to enable fitness-for-use. Metadata management to compose the context. Management of data dictionaries, privacy, tagging, hierarchies, lookups, and categorization.
3. Statistics, Probability, Forecasting, & Evaluation
4. Simulation, War-gaming, & Situation Awareness
5. Optimization, Queuing, & Game Theory
6. Decision Analysis & Decision Coaching
7. Industrial Engineering & Systems Science
8. Decision Intelligence Improvement & Reduction
   1. Decision Intelligence Improvement. This is about strengthening the ability to make rational decisions by harnessing methods and data into playbooks and decision-models that are used to generate outcomes that are desired by the user of the Decision Intelligence.
   2. Decision Intelligence Reduction. This is about reducing or eliminating a target’s ability to make decisions. Playbooks drive outcomes desired by their creators, and need not align with the best interests of the creator’s targets, for instance a marketing genius who targets consumers to buy cigarettes. This is common in advertising, e-commerce, and politics. Uses methods to reduce “friction” or add “nudge” so that people slide to an analyst-desired action (such as buying an overpriced product or supporting a candidate) without thinking deeply about the decision, where thinking may delay, impede, or bypass the analyst-desired decision.
9. Accounting, Finance, & Economics
10. Knowledge Management. The ability to distil, store, deploy, and update knowledge.

What is an intelligent organization, and how does it become more intelligent? How do we realize the dream of the Intelligent Enterprise? What is the role of a leader?

1. Transition to systematic decision intelligence. Understand that many outcomes have a large component of chance or dependence on external trends and forces. Use systematic decision intelligence to deal with the fact that many management methods do not result in making better decisions. Drive continuous improvement of the decision intelligence system.
2. Build, use, and improve playbooks. Be open to using external expertise in making internal decisions, make playbooks with expertise from external and in-house sources.
3. Understand and manage data quality. Handle the reality that data in business systems that should be directly usable for making decision is usually not ready-for-use because it takes care and effort to make data of high quality, put relentless attention and efforts into data quality improvement.