Skip to content

Explainability Framework

The core differentiator of CostEngine (MfgIQ) is not just calculating a number, but proving it. The Explainability Framework ensures every cost is traceable, defensible, and transparent.

The "Founder's Vision" for Explainability

"In manufacturing, a quote is a negotiation. If you can't explain every rupee, you lose the deal or your margin. We don't just give you a price; we give you the logic to defend it."

Three Pillars of Explainability

1. Data Lineage (The "Where?")

Every input must point to its origin. - Excel Origin: "This gross weight came from cell B14 of the uploaded sheet." - Master Data: "This RM rate was pulled from the Material Master on 2026-01-24." - Manual Input: "This cycle time was overridden by Estimator Rahul."

2. Mathematical Transparency (The "How?")

Formulas should be human-readable, not a "black box." - Instead of showing just result ₹529, show: (Material ₹264) + (Operations ₹238) + (Overhead ₹27). - Enable "Formula Drill-down" for any calculated field.

3. Change Traceability (The "Why?")

Why did the price change from v1.0 to v1.1? - Track differences in inputs (RM rate hike) vs. differences in process (added heat treat). - Compare Versions is the primary UI for this.

Explainability UI Patterns

A. The "Explain This" Modal

Available on any cost line item. It displays: 1. Inputs: Values used. 2. Formula: The actual math applied. 3. Assumptions: Hidden variables (e.g., Efficiency = 80%). 4. Overrides: Indication if the value was modified manually.

B. Inline Source Tags

Subtle badges next to data fields: - [EXCEL] Blue: From the uploaded file. - [MASTER] Green: From central database. - [CALC] Purple: Result of a formula. - [EDITED] Amber: Manually changed.

Hierarchy of Explainability

CostEngine breaks down explainability into levels:

Level Audience Goal
Executive Management High-level margin check and risk assessment.
Estimator Costing Team Detailed validation of every operation and weight.
Customer Procurement Professional breakdown to justify the price.

Technical Implementation (Metadata)

The system stores an explainability_blob for every quote version.

{
  "metric": "raw_material_cost",
  "value": 263.95,
  "logic_path": "MaterialSheet -> FormulaV1 -> Result",
  "data_points": {
    "gross_weight": {"value": 2.304, "source": "excel://Sheet1!D13"},
    "rm_rate": {"value": 99.5, "source": "db://rm_master/SAE4140"}
  },
  "formula_string": "(gross_weight * rm_rate) / yield"
}

Best Practices for Estimators

  1. Add Comments on Overrides: Always explain why you changed a machine time. "Machine #4 is running slower this week" is a defensible reason.
  2. Update RM Rates Weekly: Explainability depends on current market data.
  3. Verify Yield Labels: Ensure your "Net" and "Gross" markers in Excel align with the system's mapping.