Infrastructure for Procurement Forecasting & Capacity Planning

Capacity planning ML models require explicitly documented forecasting assumptions: how seasonal patterns are calculated, what customer growth inputs feed the model, how contracted capacity is compared against forecast demand, and what lead time thresholds trigger RFP recommendations. At L3, these model parameters and planning procedures are documented and findable, enabling the AI to produce auditable gap analyses rather than black-box forecasts that procurement managers cannot validate or trust for carrier contracting decisions.

Forecasting models require systematic capture of historical shipment volumes by lane, seasonal patterns, customer-provided growth forecasts, and contract capacity commitments. At L3, TMS workflow templates enforce consistent lane-level volume recording, and customer growth inputs are captured through defined planning templates — not ad-hoc emails. This systematic capture ensures the ML model trains on complete, consistent historical data rather than whatever was manually entered into a spreadsheet by whoever had time.

Capacity forecasting requires consistent schema linking shipment history to lanes (origin-destination pairs), equipment types, time periods, customers, and contract capacity commitments. At L3, all historical volume records share these required fields, enabling the ML model to compute seasonality indices by lane and project capacity needs against contracted capacity by the same dimensions. Gap analysis output (forecast demand minus contracted capacity by lane/month) requires this structural consistency to be computationally valid.

Procurement forecasting requires API access to TMS historical shipment data by lane, customer account systems for growth forecasts, and contract management for committed capacity balances. At L3, the forecasting model queries these systems programmatically to assemble the inputs for capacity planning — historical volumes, growth adjustments, and committed capacity — without requiring staff to manually export and combine data from multiple sources before each planning cycle.

Capacity planning accuracy depends on event-triggered updates when new customer wins are announced, when contract capacity changes are executed, or when market capacity signals shift significantly. At L3, a new customer contract triggers an update to the capacity planning model's demand inputs, and contract amendments trigger updates to committed capacity baselines. This keeps the gap analysis current enough to drive actionable RFP timing recommendations rather than plans built on stale assumptions.

Procurement forecasting requires API-based connections between TMS (historical shipment volumes by lane), CRM or sales system (customer growth forecasts), contract management (committed capacity), and the planning output system (RFP roadmap). At L3, these systems are connected via APIs so the forecasting model assembles inputs programmatically and outputs gap analyses and RFP timing recommendations that procurement can act on directly, rather than manually reconciling data across disconnected spreadsheets.