Skip to main content

What-If Workbench

The What-If Scenario Workbench​

Version: 2.0 Date: September 8, 2025 (Revised based on Executive UX Feedback)

1. Feature Overview & Objective​

The What-If Scenario Workbench is a strategic decision tool designed to provide a collaborative, real-time environment for the S&OP team to model and understand the full business impact of their choices. Its primary objective is to lead with conclusions and minimize cognitive load, allowing executives to make confident, data-driven trade-off decisions in seconds.

2. Core Design Principles​

  • Lead with the Conclusion: Every scenario view must start with a clear, business-language headline that summarizes the primary insight and trade-off.

  • Minimize Cognitive Load: The default view will always be a simple comparison of a single scenario against the baseline. A full matrix of all scenarios is a secondary, drill-down option.

  • Model Outcomes, Not Just Tactics: The interface for creating scenarios will focus on strategic business outcomes (e.g., "Eliminate stockout risk") rather than granular operational levers.

  • Quantify Risk, Don't Just Show Data: All probabilistic simulations must be presented with clear, plain-language confidence levels and business conclusions.

3. Refined User Workflow​

  1. Initiate Scenario: The user initiates a new scenario from the main dashboard.

  2. Define Strategic Objective: The user interacts with the new "Strategic Scenario Builder," selecting a desired business outcome or setting custom targets.

  3. Review Impact Analysis: The system generates the "Decision-Focused Layout," presenting the results of the new scenario against the baseline with a clear "Primary Insight" headline.

  4. Assess Risk (Optional Deep Dive): The user can trigger a Monte Carlo simulation. The results are displayed in the "Enhanced Probabilistic Results" view, which leads with business conclusions.

  5. Promote & Log Decision: The facilitator promotes the chosen scenario to become the new "Recommended Plan," triggering the "Consensus Lock-In Protocol".

4. Detailed Functional Specifications of UI Components​

4.1. The Redesigned Main View: "Decision-Focused Layout"​

This is the default view for analyzing a single scenario against the baseline, designed to pass the "10-second rule."

  • a. AI-Generated "Primary Insight" Headline: The main title of the view will be an AI-generated, McKinsey-style conclusion that summarizes the core trade-off.

Example: "Trading $1.5M in working capital for a 14-point stockout risk reduction delivers a net $2.8M benefit."

  • b. Key Impact Table: A simple, high-contrast table showing only the most critical KPIs for the "Baseline" vs. the "Scenario."

| KEY IMPACT | BASELINE | EXPEDITE EUR |

| :--- | :--- | :--- |

| Revenue at Risk | $4.2M | $1.4M |

| Working Capital | $22M | $23.5M |

| Service Level | 95% | 98% |

  • c. Actions: The view will contain clear action buttons like [Run Monte Carlo Simulation] and [Promote to Plan]. An expandable [View All Scenarios] button will be available for deeper analysis during reconciliation meetings.

4.2. The New "Strategic Scenario Builder"​

This modal replaces the granular operational levers with business outcome choices.

  • a. Strategic Options: The user is first presented with a set of pre-defined strategic goals, derived from the current business problem.

  • ○ Eliminate Q1 stockout risk (Accept ~$1.5M cost)

  • ○ Maximize gross margin (Accept ~12% stockout risk)

  • ○ Maintain current service level (Minimize cost)

  • b. Custom Scenario: Allows power users to set specific targets for key metrics.

  • Target Service Level: [ 99 ]%

  • Working Capital Limit: $[ 25 ]M

  • c. Action: The [Generate Impact Analysis] button triggers the backend to run the necessary calculations and present the results in the "Decision-Focused Layout."

4.3. The Enhanced Probabilistic Results View​

When a Monte Carlo simulation is run, the results are presented in a new card that leads with business conclusions.

  • a. Headline Conclusion: The primary output is a plain-language summary of the risk profile.

"78% probability of achieving service targets with a 15% chance of exceeding cost estimates."

  • b. Confidence Levels: The view translates the P10/P50/P90 distribution into clear, understandable confidence statements.

  • • 90% certain: Revenue protection will be greater than $2.1M.

  • • 50% likely: Total cost will be under $1.8M.

  • • 10% risk: Cost overrun could exceed $2.5M.

  • c. Actions: Buttons like [View Full Distribution] (for the detailed chart) and [Accept Risk Profile] will be available.

4.4. Integration with Design System & Meeting Flow​

  • a. Color Strategy: As specified in the feedback, Signal Green (#00D084) will be used for positive impacts and recommended actions. Amber (#F59E0B) will be used for risk indicators. Deep Navy (#0B1D3A) will be used for baseline data.

  • b. Adaptive UI: The workbench will adapt its complexity based on the meeting phase.

  • Supply Reconciliation: Full workbench access with the "View All Scenarios" matrix may be the default.

  • Executive S&OP: The simplified, single-comparison "Decision-Focused Layout" will be the default view.

Notes​

Use of Streaming​

How Streaming Powers the Workbench Workflow​

Here are the two primary moments in the user workflow where streaming should be implemented:

1. Generating the Impact Analysis​

When a user defines a scenario in the "Strategic Scenario Builder" and clicks [Generate Impact Analysis], a complex series of calculations is triggered.

Without Streaming: The user sees a loading spinner for several seconds before the entire "Decision-Focused Layout" appears.

With Streaming: The experience becomes dynamic and immediate:

  1. Instantly Load the Shell: The system immediately renders the empty "Decision-Focused Layout".

  2. Stream the Headline First: The backend runs the LLM call for the AI-generated "Primary Insight" headline. As soon as this is ready (often in 1-2 seconds), it's streamed to the UI. The user immediately gets the key takeaway, fulfilling the "Lead with the Conclusion" principle.

  3. Populate KPIs Progressively: As the backend calculates each metric for the Key Impact Table, it streams them one by one. The user might see "Revenue at Risk" populate first, followed by "Working Capital," and then "Service Level". This creates a sense of live progress.

2. Running a Monte Carlo Simulation​

This is a computationally intensive "deep dive". Streaming is crucial for making this feature usable without a long, frustrating wait.

Without Streaming: The user clicks [Run Monte Carlo Simulation] and waits, potentially for 10-30 seconds, before any results appear.

With Streaming: The "Enhanced Probabilistic Results" view comes alive instantly:

  1. Stream the Headline Conclusion: A preliminary "Headline Conclusion" can be generated and streamed based on the first pass of the simulation.

  2. Stream Confidence Levels as They Stabilize: The simulation calculates different probability levels (P10, P50, P90). The system can stream these values as they reach statistical significance. For instance, the "90% certain" value might stabilize and appear first, followed by the "50% likely" value, and finally the riskier "10% risk" value.

Summary of Streaming Logic​

User ActionSystem Response (with Streaming)
Clicks [Generate Impact Analysis]Instantly show the layout, stream the AI headline first, then progressively populate the KPI table.
Clicks [Run Monte Carlo Simulation]Instantly show the results card, stream the headline conclusion, then stream each confidence level as it is calculated.

By implementing streaming this way, you transform the workbench from a static, request-and-wait tool into the dynamic, live decision-making environment envisioned in your functional specification.