Inventory demand forecasting

Requirements describe capabilities the pattern needs in your environment, not the vendors you must buy. Any system that fills a requirement satisfies it — that’s what makes the catalog portable across the long tail of SMB tooling.

1. sales_history_corpus

   REQUIREDREADbatch

   Historical sales data at the granularity that drives forecasting: per SKU, per location, per time period.

   DATA SHAPE

   Sales records with SKU, location, quantity, date, sale price, optionally customer segment.
   COMMONLY FILLED BY

   • sales transactions from the e-commerce platform
   • POS data from retail systems
   • ERP sales orders
   • subscription billing data for recurring products
2. inventory_state

   REQUIREDREADrequest

   Current and historical inventory positions: what's on hand, what's on order, what's in transit.

   DATA SHAPE

   Per SKU per location: on-hand quantity, allocated quantity, open POs with expected arrival.
   COMMONLY FILLED BY

   • inventory management system
   • ERP inventory module
   • warehouse management system
3. product_master

   REQUIREDREADrequest

   SKU-level information that drives forecasting and reordering: lead time from supplier, MOQ, shelf life, cost.

   DATA SHAPE

   Per-SKU: supplier, lead time, MOQ, unit cost, ABC classification, lifecycle stage.
   COMMONLY FILLED BY

   • product catalog in ERP
   • purchasing system with supplier and lead-time data
   • master product list maintained by operations
4. external_signal_inputs

   RECOMMENDEDREADcorpus

   Things that affect demand but aren't visible in historical sales alone: known promotions, weather, holidays, planned events.

   DATA SHAPE

   Calendar of events with date ranges and expected demand multipliers (positive or negative) per SKU group.
   IF MISSING

   Forecasts work well for steady-state demand but miss promo bumps and seasonal anomalies. Accuracy 15-25% worse on event-driven products.
   COMMONLY FILLED BY

   • marketing calendar with planned campaigns
   • annual events the business runs
   • weather feed for weather-sensitive products
5. recommendation_destination

   REQUIREDWRITEbatch

   Where reorder recommendations land for the buying team to review and act on.

   DATA SHAPE

   Per SKU recommendation: order quantity, suggested order date, forecast supporting it, confidence range, alternative scenarios.
   COMMONLY FILLED BY

   • review queue in the purchasing module
   • weekly recommendation report
   • dashboard with sortable recommendations
6. buying_outcome_loop

   RECOMMENDEDREADevent

   Feedback on which recommendations the buyer accepted, modified, or rejected. Used to tune forecasts.

   DATA SHAPE

   Per-recommendation: was it acted on, what was actually ordered, what was the buyer's reasoning if different.
   IF MISSING

   Pattern can't learn from buyer expertise. Recommend at least minimal capture of accept/reject.
   COMMONLY FILLED BY

   • accept/reject log captured at recommendation time
   • PO data fed back from the purchasing system
   • weekly buyer feedback meeting captured in notes