Six Layers. One Source of Truth.
The Guidance architecture is organized into six sequential layers. Each layer takes the output of the previous one as its input. In the commercial design, this produces a connected calculation path from raw ingredient cost to channel-level margin, with confidence attached to every material input.
Raw Data In. Clean Data Out.
Before any math can happen, the data has to be clean. The commercial design is intended to ingest supplier invoices, distributor remittance files, production records, and sales data and normalize them into a unified internal format.
The intended scope includes distributor deduction normalization. UNFI, KeHE, and other distributors use different taxonomies for deductions such as spoils, MCB, OI, EDLC, and scan-backs. The commercial design is meant to map those records to a common taxonomy so deductions can be categorized and attributed to the relevant SKU and promotion.
Every Ingredient. Every Overhead. One Landed Cost.
The cost-foundation design is intended to calculate landed cost for ingredients and packaging components, including invoice price, freight, duty, tariffs, and applicable Section 301 surcharges.
The intended commercial workflow connects lot-level cost records to production usage so differently priced receipts can be attributed to the relevant batch rather than averaged together.
One Price Change. Connected Updates Across Every SKU.
The prototype validated the underlying cost-propagation model. The commercial design is being rebuilt so an ingredient price change can flow through connected BOMs and identify the COGS impact on affected SKUs.
The intended result is a controlled propagation workflow that reduces disconnected spreadsheet updates and makes the effect of a reformulation or price change visible before it becomes a margin surprise.
Multi-level BOM propagation is part of the commercial rebuild scope so a component-cost change can flow through sub-assemblies to affected finished goods.
COGS That Reflects What Actually Happened in Production.
Standard COGS calculations reflect what a team expected to spend. The commercial design is intended to calculate lot-level actual COGS using production evidence such as yield variance, rework, waste, and recorded deviations.
In the intended workflow, a 94% actual yield versus a 97% expected yield would change the attributed cost for that lot. Deviation detection and historical yield ranges remain commercial rebuild scope rather than generally available functionality.
True Net Margin. Per SKU. Per Channel. After Everything.
COGS is only half the picture. What a brand actually keeps depends on which channel the product was sold through. A unit sold through UNFI has a completely different net margin than the same unit sold DTC or on Amazon, because each channel has its own deductions, fees, freight costs, and promotional allowances.
The channel-margin design is intended to combine SKU cost with channel-specific deductions and fees, including distributor deductions, Amazon fees, and DTC transaction and fulfillment costs. This remains commercial rebuild scope; integrations and production reconciliation are not yet generally available.
Every Number Comes With a Grade.
The prototype validated a Confidence Score model from 0 to 100 for distinguishing confirmed inputs from estimates and missing data. The commercial rebuild is intended to preserve that evidence trail across costing and margin calculations.
This matters because CPG teams routinely make decisions with incomplete data. The intended model shows which inputs are confirmed and which require review before a pricing or channel-allocation decision.
The planned optimization layer would use confidence-scored data to surface potential margin deterioration, channel underperformance, and scenarios for price or allocation changes. These recommendation workflows are not yet generally available.