The Decision Intelligence Process: 5 Phases for integrating AI into Decision-Making

The Decision Intelligence Process: Handbook

The Decision Intelligence (DI) Handbook by Lorien Pratt and N. E. Malcolm introduces a practical framework for improved decision-making, centred around the Causal Decision Diagram (CDD). The CDD, a visual tool, maps actions to outcomes, highlighting intermediate steps, dependencies, and external factors. DI prioritises understanding the cause-and-effect relationships between actions and measurable outcomes, differentiating itself from simpler decision trees. The handbook stresses the iterative, collaborative nature of DI, incorporating existing knowledge, simulating outcomes, and monitoring results, aligning with the Observe-Orient-Decide-Act (OODA) loop. Crucially, while technology can enhance DI, its core principles remain applicable even without sophisticated software, emphasising human expertise and structured thinking.

Main Themes

• Decision Intelligence (DI) as a Practical Framework: The handbook emphasizes the immediate applicability of DI, encouraging readers to start building their first Causal Decision Diagram (CDD) within minutes.
• CDD as the Core of DI: The CDD serves as a visual blueprint for decisions, mapping actions to outcomes through intermediates and dependencies. It facilitates shared understanding and structured decision-making.
• DI Beyond Technology: While technology can enhance DI, its core principles can be applied even without sophisticated tools. Human knowledge and collaborative processes are fundamental.
• Iterative and Collaborative Nature of DI: The handbook highlights the importance of iteratively refining CDDs and actively involving stakeholders in the process.
• DI Maturity Model: The book introduces a maturity model, outlining how organizations can progressively adopt and enhance their DI capabilities.

Most Important Ideas/Facts

• Definition of DI: “DI helps organizations make better decisions. It helps decision makers understand how the potential actions they can take today (the things they can do) could affect their desired outcomes (the things they want to accomplish).”
• Focus on Action-to-Outcome Decisions: DI specifically targets decisions where actions lead to measurable outcomes, as opposed to process-oriented or data-driven conclusions.
• Distinction from Decision Trees: DI goes beyond sequential questioning by explicitly modelling cause-and-effect relationships between actions and outcomes.
• Key CDD Elements:Objectives: Measurable targets for desired outcomes.
• Levers: Sets of choices leading to actions that influence outcomes.
• Intermediates: Steps along the causal chain between actions and outcomes.
• Dependencies: Relationships illustrating how changes in one element affect others.
• Externals: Factors outside the decision-maker’s control that can impact outcomes.
• OODA Loop Integration: DI aligns with the Observe-Orient-Decide-Act (OODA) loop, providing a framework for iteratively making decisions over time.
• Decision Asset Investigation: Identifying existing data, models, and knowledge relevant to the decision and prioritizing new data gathering or model building.
• Decision Simulation: Creating a “digital twin” of the CDD to simulate how actions lead to outcomes and identify optimal choices.
• Decision Assessment: Evaluating the decision model’s fidelity, accuracy, risk, and uncertainty through various lenses.
• Decision Monitoring: Tracking key elements after taking action to identify deviations from expectations and trigger necessary adjustments.

Key Quotes

• “DI is a powerful approach even when you use no computer support for making the decision or for tracking the decision after it’s made (beyond common office tools like Microsoft Office or Google Workspace).”
• “Unlike other mathematical methods for decision formalization, the most important criterion here is to maximize the cognitive “space” available to the customer to think about their situation.”
• “You can think of these time phases as playing out in an observe-orient-decide-act (OODA) loop… Indeed, we might think of DI as showing us how to do OODA loops when we have a computer to help us to reason.”
• “Fidelity means the degree to which your model and reality match. A model that perfectly matches reality will—by definition—produce good decisions. But the opposite isn’t true: as we noted earlier in this chapter, a decision model can have low fidelity and still reliably lead to accurate decisions.”

Overall Impression

The Decision Intelligence Handbook provides a comprehensive and accessible introduction to the DI framework. It emphasizes the practical application of DI principles, empowering individuals and organizations to make better, more informed decisions through structured thinking, collaboration, and the effective utilization of available data and knowledge. The emphasis on the iterative nature of the process and the integration of human expertise ensures that DI remains a human-centered approach, even when leveraging technology.

5 Phase DI Methodology (9 Steps)

The text outlines a nine-step Decision Intelligence (DI) methodology for making complex decisions. This methodology is structured across five phases: defining the decision’s objective and scope (Phase A), building a decision model (Phase B), using the model for simulation and assessment (Phase C), monitoring the implemented decision (Phase D), and finally reviewing and retaining the process’s learnings (Phase E). The overall purpose is to provide a structured framework for data-driven decision-making, improving the effectiveness and informing of choices.

Phase A: Decision Requirements

• A1. Decision Objective Statement: This initial step involves creating a clear and concise statement that outlines the decision to be made. It’s more of a trigger to initiate the DI process, setting the scope and context for subsequent steps.
• A2. Decision Framing: This process defines the boundaries and constraints of the decision. It clarifies what is in and out of scope, identifies desired outcomes and available actions, and verifies the suitability of DI for the decision at hand.

Phase B: Decision Modelling

• B1. Decision Design: This is where the visual blueprint for the decision, the CDD, is created. The team collaborates to identify outcomes, levers (actions), externals (factors beyond control), and the causal chains connecting them.
• B2. Decision Asset Investigation: This process involves identifying existing data, models, and expertise (decision assets) that can inform the decision. The CDD is annotated to reflect these assets and highlight any missing information that needs to be gathered.

Phase C: Decision Reasoning

• C1. Decision Simulation: Here, the CDD is used to simulate different scenarios and understand how various actions and external factors might impact the desired outcomes. This helps to identify optimal choices and potential risks.
• C2. Decision Assessment: This process evaluates the decision model’s fidelity, accuracy, risk, and uncertainty. It involves examining the model from different perspectives (lenses) to ensure its reliability and identify potential areas for improvement.

Phase D: Decision Action

• D1. Decision Monitoring: This process involves tracking the results of the decision after action has been taken. It ensures that the decision is on track, allows for adjustments if deviations occur, and helps to determine if the decision has become obsolete.

Phase E: Decision Review

• E1. Decision Artifacts Retention: This process involves preserving all the information and documentation generated throughout the DI process. This creates a valuable repository of knowledge that can be reused for future decisions.
• E2. Decision Retrospective: This final process involves reflecting on the decision-making experience and identifying lessons learned. It assesses the effectiveness of the process, identifies areas for improvement, and contributes to a culture of continuous improvement.

These nine processes provide a structured and iterative framework for navigating complex decisions, leveraging data, expertise, and simulation to make more informed and effective choices.

Phase A: Decision Requirements

Phase A: Decision Requirements is the first phase in the Decision Intelligence (DI) methodology and it sets the stage for subsequent DI processes, aligning the team around the decision to be made. It consists of two processes:

A1. Decision Objective Statement:

• This initial step involves creating a clear and concise statement that outlines the decision to be made. This statement acts as a trigger to initiate the DI process and sets the scope and context for subsequent steps.
• It should articulate the core decision to be made, capturing the essence of the problem or opportunity at hand. The Decision Objective Statement should be thought of as a ‘north star’ guiding the decision-making journey.
• The Decision Objective Statement doesn’t need to delve into specifics or solutions at this stage. It is more of a starting point, an imprecise, unrefined description of the decision in the decision maker’s own words.

A2. Decision Framing:

• This process further defines the boundaries and constraints of the decision. It clarifies what is in and out of scope, identifies desired outcomes and available actions, and verifies the suitability of DI for the decision at hand. Decision Framing seeks to establish a common understanding of all the decision requirements and deliverables between the customer and the team. This is achieved by establishing and agreeing to the ‘non-negotiable guardrails’ of the decision-making process.
• During Decision Framing, the team will confirm that the Decision Objective Statement represents a decision that is appropriate for DI. DI is specifically for decisions where actions lead to measurable outcomes. This is called decision verification. The team also needs to consider how similar decisions were made in the past to establish a decision baseline. This helps identify areas where DI can add value and streamline the decision-making process. The Decision Framing process produces a Decision Framing Worksheet that documents the frame of the decision, including any constraints, boundaries, and/or requirements that come from outside of the decision team
• It is tempting to skip Phase A but this should be avoided because starting off on the wrong track can limit the value of all downstream processes.

Phase B: Decision Modelling

B1. Decision Design:

• This is where the visual blueprint for the decision, the Causal Decision Diagram (CDD), is created. The team collaborates to identify outcomes, levers (actions), externals (factors beyond control), and the causal chains connecting them.
• This process focuses on aligning around the outcomes, identifying levers that can produce those outcomes, understanding externals that influence outcomes, and building the causal chains from levers and externals to outcomes.
• The goal is to get everyone on the same page – technologists, decision makers, and even the stakeholders affected by the decision.

B2. Decision Asset Investigation:

• This process focuses on identifying the assets that will ultimately help to implement those causal chains in software.
• This process involves identifying existing data, models, and expertise (decision assets) that can inform the decision.
• The CDD is annotated to reflect these assets and highlight any missing information that needs to be gathered.
• This transforms your ‘decision blueprint’ into a ‘decision digital twin‘ specification, ready for simulation.

Completing Phase B ensures that the decision-makers, stakeholders, and the technology teams are aligned around a shared understanding of the causal chains that lead from actions to outcomes. This common understanding helps the team identify evidence that can inform the decision

Phase C: Decision Reasoning

Decision Reasoning (Phase C) in Decision Intelligence (DI) is where you take your Causal Decision Diagram (CDD), along with the assets you’ve identified, and explore how different actions and external factors might affect your desired outcomes. Think of it as bringing your decision to life – no longer just a static diagram, but a dynamic system where you can experiment and test different scenarios. This helps you to identify both optimal choices and potential risks, before you commit to a course of action. Phase C consists of two processes:

C1. Decision Simulation:

• This involves creating a ‘digital twin’ of your CDD, allowing you to simulate different scenarios and see how changes to levers (actions) and externals (factors outside of your control) impact the outcomes. Imagine it as a video game where you try different combinations to see what “scores” the best results. There are various ways to simulate your CDD, ranging from simple mental simulations to sophisticated software simulations. The level of sophistication you choose depends on factors like the complexity of the decision, the availability of data and models, and the desired level of accuracy.
• The benefits of Decision Simulation include:
  • A clearer understanding of cause and effect.
  • Identification of optimal actions.
  • Early detection of potential risks and unintended consequences.

C2. Decision Assessment:

• Once you have simulated your CDD, it’s time to evaluate the model itself. This process involves examining the model from different perspectives, or “lenses”, to assess its reliability and identify potential areas for improvement. You are essentially checking the trustworthiness of your “decision map” before making any real-world decisions.
• Key Decision Assessment lenses include:
  • Accuracy.
  • Bias.
  • Sensitivity.
  • Fidelity.
  • Uncertainty.
• By examining the model through these lenses, you can identify any weaknesses or areas where the model might be oversimplifying or misrepresenting reality. This allows you to make adjustments to the model or gather additional information before making your final decision.

The takeaway from Phase C is that you are not just building a model, you are learning from it. By simulating and assessing your CDD, you gain valuable insights into the dynamics of your decision, the potential consequences of your actions, and the strengths and weaknesses of your model. This knowledge empowers you to make more informed and confident decisions, with a greater understanding of both the risks and the opportunities involved.

Phase D: Decision Action

Decision Action (Phase D) in Decision Intelligence (DI) marks the transition from planning and analysis to real-world implementation. This phase is where the insights gleaned from the previous phases – framing, modelling, and reasoning – are translated into tangible actions. However, the process doesn’t end with taking action. DI emphasises the importance of continuous monitoring to ensure the decision remains effective as the situation evolves.

Phase D consists of a single, yet crucial process:

D1. Decision Monitoring:

• This involves systematically tracking the results of the decision after action has been taken. It’s akin to using a navigation system, not just to plan a route but also to guide you along the way, alerting you to traffic changes or road closures that might require adjustments to your original plan.
• The Role of Decision Monitoring
  • Keeping the Decision on Track: By tracking key metrics (outcomes, intermediates, and externals), you can quickly identify any deviations from the expected results. This allows for timely intervention and course correction, preventing the decision from veering off course.
  • Enabling Agile Adjustments: Real-world situations are rarely static. Market conditions change, new competitors emerge, and unforeseen events occur. Decision Monitoring helps you adapt to these changes by providing the data you need to reassess your assumptions and make necessary adjustments to your actions or even the decision model itself.
  • Detecting Decision Obsolescence: Sometimes, a decision that was once sound becomes outdated or irrelevant due to significant shifts in the environment. Decision Monitoring helps you recognise when a decision has reached its “expiry date”, prompting a review and potential revision or replacement.
• Best Practices for Effective Decision Monitoring
  • Identify Key Monitoring Elements: Before taking action, carefully consider which elements of your decision model (levers, outcomes, intermediates, and externals) are most critical to track. Focus on those elements that have the most significant impact on the desired outcomes or are most susceptible to change.
  • Establish Monitoring Intervals: Determine the appropriate frequency for tracking the chosen elements. Some elements might require daily monitoring, while others might only need weekly or monthly updates. The frequency should align with the pace of change in the environment and the potential impact of deviations.
  • Define Acceptable Ranges: For each monitored element, establish “safe” or acceptable ranges based on your analysis and understanding of the system. Deviations outside these ranges should trigger alerts and prompt further investigation.
  • Develop a Monitoring Plan: Document your monitoring strategy, including the elements to be tracked, monitoring intervals, acceptable ranges, and the individuals responsible for collecting and analysing the data.
  • Leverage Monitoring Tools: Take advantage of tools that can streamline the data collection and analysis process. Dashboards, automated reports, and alert systems can significantly enhance the efficiency and effectiveness of Decision Monitoring.
• Integrating Decision Monitoring with Earlier Phases: Decision Monitoring isn’t an isolated activity; it’s closely intertwined with the insights and outputs from the preceding phases.
  • Decision Framing (Phase A): The defined scope, objectives, and constraints provide the context for monitoring, ensuring that the focus remains on the most relevant aspects of the decision.
  • Decision Modelling (Phase B): The CDD serves as a visual guide for identifying the key elements to monitor and understanding the causal relationships between them.
  • Decision Reasoning (Phase C): The simulations and assessments performed in this phase help to determine the sensitivity of outcomes to various factors, informing the selection of monitoring elements and the definition of acceptable ranges.

Decision Monitoring is an essential component of the DI methodology, ensuring that decisions don’t just remain static plans but adapt and evolve in response to real-world dynamics. By embracing Decision Monitoring, organisations can move beyond simply making decisions to actively managing them, maximising their effectiveness and achieving sustainable success.

Phase E: Decision Review

One of the great benefits of Decision Intelligence (DI) is that it allows organisations to learn from their decision-making experiences and continuously improve their decision-making processes. This is where Phase E, Decision Review, comes into play. Decision Review is about capturing the valuable knowledge generated throughout the DI process and using it to inform future decisions. Think of it as building an organisational “decision library,” where the insights from past decisions can be accessed and reused, much like engineers refer to blueprints and design documents.

Phase E consists of two key processes:

E1. Decision Artefacts Retention:

• This process involves systematically preserving all the information and documentation generated throughout the DI process. This includes everything from the initial Decision Objective Statement and Decision Framing Worksheet to the CDD, simulation results, assessment reports, and monitoring data. These artefacts are valuable sources of knowledge that can be reused for future decisions, saving time and effort, and helping to avoid repeating past mistakes.
• Key benefits of Decision Artefacts Retention include:
  • Building a repository of knowledge: By preserving decision artefacts, organisations create a valuable knowledge base that can be accessed and leveraged for future decision-making initiatives.
  • Facilitating reuse and knowledge transfer: Captured artefacts can serve as templates, examples, or starting points for new decisions, streamlining the process and fostering consistency in decision-making practices.
  • Enabling continuous improvement: By analysing past decisions, organisations can identify patterns, trends, and areas for improvement in their decision-making processes, leading to better decisions over time.

E2. Decision Retrospective:

• This process involves taking a step back and reflecting on the entire decision-making experience, from the initial framing to the final outcomes. The goal is to identify lessons learned, both positive and negative, and to use those insights to improve future decision-making processes. This involves asking questions such as:
• Did we achieve the desired outcomes? If not, why not?
• Was the decision-making process effective? What could we have done better?
• Were there any unforeseen challenges or surprises? How did we handle them?
• What new knowledge or insights did we gain through this process?
• The Decision Retrospective should be conducted as a collaborative exercise, involving all key stakeholders in the decision-making process. This ensures that diverse perspectives are captured and that everyone has the opportunity to contribute to the learning process. The insights gained from the retrospective are then used to update organizational best practices, refine decision-making frameworks, and improve the training of decision-makers.

Measuring Decision Quality

A crucial aspect of Decision Review is assessing the quality of the decision itself. This involves examining how the decision played out in the real world and comparing the actual outcomes to the desired objectives. This can be done in various ways, depending on the type of decision and the availability of data:

• One-Off Decisions with Known Outcomes: For decisions with clearly defined outcomes, the quality can be measured by comparing the actual results to the predicted or desired outcomes. For example, if the decision was to launch a new product and the sales figures fell short of expectations, the retrospective would explore the reasons for the discrepancy.
• Repeated Decisions: For decisions that are made regularly, the focus is on identifying trends and patterns over time. By analysing the outcomes of multiple instances of the same decision, organisations can identify areas for improvement and refine their decision-making processes to achieve more consistent results.
• Decisions without Clear Outcomes: For decisions where the outcomes are less tangible or take longer to materialise, the assessment might focus on the quality of the decision-making process itself. This could involve evaluating factors such as the clarity of objectives, the thoroughness of analysis, the effectiveness of communication, and the level of stakeholder engagement.

Building a Culture of Continuous Improvement

Phase E, Decision Review, is not simply about evaluating past decisions; it’s about embedding a culture of continuous improvement in the organisation’s DNA. By systematically capturing lessons learned and using them to refine decision-making processes, organisations can create a virtuous cycle of learning and improvement. Decision Review is crucial for ensuring that DI delivers on its promise of better, more informed, and more effective decisions.

Key takeaways for effective Decision Review:

• Start early: Don’t wait until the end of the DI process to start thinking about Decision Review. Begin capturing information and documenting decisions from the outset.
• Make it a collaborative process: Involve all key stakeholders in the decision retrospective to gather diverse perspectives and ensure buy-in for any proposed improvements.
• Focus on learning, not blame: The goal is to identify areas for improvement, not to assign blame for past mistakes. Create a safe environment for open and honest discussion.
• Document and share lessons learned: Make the insights gained from the Decision Retrospective accessible to everyone in the organization, ensuring that knowledge is shared and that future decisions benefit from past experience.

By embracing Decision Review, organisations can move beyond simply making decisions to becoming learning organisations that continuously adapt and evolve in response to new information and changing circumstances. This is the essence of Decision Intelligence – a journey of continuous improvement and a commitment to making the best decisions possible, both now and in the future.convert_to_textConvert to source