Reference

Retail AI Glossary

Definitions for the key concepts behind retail decision platforms — allocation, replenishment, markdown, forecasting, supply chain and financial terms.

Decision platform

Allocation
Distributing stock across stores before the selling season starts to match expected local demand.Read more →
Decision layer
The software layer between raw data and execution systems. It translates inventory positions into actionable allocation, replenishment and markdown orders.Read more →
Decision loop
The cycle from data collection to decision to execution to feedback. Closing this loop is what separates high-performing retailers.Read more →
Markdown
A planned price reduction applied to clear inventory before the end of a selling season.Read more →
Markdown depth
The magnitude of a price reduction, typically expressed as a percentage of the original selling price.
Open-to-buy
The remaining purchasing budget available for a category within a given period, after committed orders.
Override
A manual revision of an automated decision made by a buyer or planner.
Override rate
The proportion of automated decisions that are manually revised. A high rate signals model miscalibration or cultural resistance to the platform.
Replenishment
Restocking a store from a distribution centre to maintain target inventory levels.Read more →
Sell-through rate
The percentage of received inventory sold within a defined period. A key input for markdown and replenishment decisions.
Shadow mode
Running automated decisions in parallel with manual processes without acting on them, to validate model accuracy before full deployment.

AI & Data

AI agent
An autonomous software component that perceives state, selects actions and executes them to achieve a goal — without human input for each step.Read more →
Censored demand
Actual customer demand not observed because the product was out of stock. Models must correct for this to avoid underestimating true demand.Read more →
Feature
An input variable used by a machine learning model to generate predictions — for example week-of-year, store tier or historical sell-through.
Holdout test
A validation methodology that withholds a subset of stores or SKUs from automation to measure incremental impact against a control group.
Lakehouse
A unified data architecture combining the flexibility of a data lake with the structure and performance of a data warehouse.
MAPE
Mean Absolute Percentage Error — the average absolute percentage deviation between forecast and actual values. A common but imperfect forecast accuracy metric.Read more →
Retraining
Updating a machine learning model on new data to maintain accuracy as demand patterns shift over time.
RMSE
Root Mean Square Error — a forecast accuracy metric that penalises large errors more heavily than MAPE.
WAPE
Weighted Absolute Percentage Error — a variant of MAPE that weights errors by actual volume. More relevant for retail than simple MAPE.

Supply chain & Ops

Distribution centre
A central warehouse that receives goods from suppliers and distributes them to stores or fulfilment points.
ERP
Enterprise Resource Planning — integrated software managing transactions, finance and master data. The system of record for inventory and orders.
Fill rate
The proportion of customer demand fulfilled from available stock on the first request, without backorders or substitutions.
Lead time
The elapsed time between placing a purchase order and receiving the goods. A key constraint in replenishment optimisation.
OMS
Order Management System — software handling order routing, fulfilment orchestration and customer-facing order visibility.
Overstock
Excess inventory beyond the optimal holding level, tying up working capital and increasing markdown risk at season end.Read more →
Safety stock
Buffer inventory held to absorb demand uncertainty and supply variability, preventing stockouts between replenishment cycles.
SKU
Stock Keeping Unit — a unique identifier for a specific product variant such as size, colour or style. The atomic unit of inventory management.
Slotting
Optimised assignment of products to warehouse locations based on picking frequency, weight and co-location logic.
Stock transfer
Movement of inventory between stores or between a DC and a store to rebalance stock without new procurement.Read more →
Stockout
An out-of-stock event where a product is unavailable at the point of sale. Causes lost sales and, if chronic, erodes customer loyalty.Read more →
WMS
Warehouse Management System — software governing warehouse operations including receiving, slotting, picking and putaway.

Finance

CapEx
Capital Expenditure — one-time investment costs such as servers, outright licences or implementation fees. SaaS decision platforms typically carry minimal CapEx.
Gross margin
Revenue minus cost of goods sold, expressed as a percentage of revenue. The primary financial metric affected by allocation, replenishment and markdown decisions.Read more →
IRR
Internal Rate of Return — the discount rate at which the net present value of an investment equals zero. Used to compare decision platform returns against other capital allocations.
OpEx
Operational Expenditure — recurring costs such as SaaS subscriptions, cloud compute and team time. Most decision platform costs are OpEx.
P&L
Profit & Loss statement — the income statement summarising revenues, costs and profit over a period. Decision platform impact is measured as a named P&L line.
Payback period
The time required for cumulative returns to recover the initial investment. Typical decision platform payback is six to eighteen months.
ROI
Return on Investment — the ratio of net benefit to investment cost, expressed as a percentage. For decision platforms, typically measured over a 24-month horizon.Read more →
TCO
Total Cost of Ownership — the full cost of a platform over its lifetime, including integration, maintenance, retraining and team time. The key metric in build-vs-buy analysis.Read more →