A retail store intelligent replenishment algorithm and process design

Abstract

The application is a retail store intelligent replenishment algorithm and process design, which is an important design and implementation basis for retail store inventory management and supply chain management, aims to improve the replenishment quantity accuracy and the replenishment process efficiency through big data integration and analysis, advanced prediction model and replenishment optimization algorithm, automatic process flow and other technologies, and finally optimizes the inventory turnover rate. The application relates to an optimized replenishment algorithm and a replenishment process of a retail store, and is mainly suitable for supermarkets, convenience stores, specialized stores and other retail sub-sectors.

Description

Technical Field

[0001] This invention relates to the field of retail and supply chain technology, specifically to functions such as big data collection, processing and analysis, sales forecasting, replenishment suggestion calculation, automatic replenishment order generation, and automatic process flow. It discloses an algorithm and process design that utilizes big data integration and analysis, advanced forecasting models, and replenishment optimization algorithms to accurately predict future sales, automatically calculate suggested replenishment quantities, automatically generate replenishment orders, and automate the process flow. Background Technology

[0002] Traditional replenishment methods rely on stores collecting basic data from various systems, such as inventory, sales, purchasing, customer traffic, and promotional plans. Replenishment quantities are then estimated using spreadsheets or personal experience, replenishment orders are created, quantities are manually entered, and the process is submitted and approved. This approach suffers from insufficient accuracy in replenishment quantities and inefficiency in the replenishment process. Therefore, a method is needed that leverages big data integration and analysis, advanced predictive models, and replenishment optimization algorithms to accurately predict future sales and iteratively self-correct. Based on current inventory and in-transit stock, and considering various influencing factors such as promotions and traffic, the replenishment optimization algorithm automatically calculates the recommended replenishment quantity, automatically generates replenishment orders, and automatically processes them. This reduces individual differences among users, improves the accuracy of replenishment quantities, increases the efficiency of the replenishment process, and ultimately optimizes inventory turnover. Summary of the Invention

[0003] This invention is a smart replenishment algorithm and process design for retail stores. It is an important design and implementation foundation for retail store inventory management and supply chain management. It aims to improve the accuracy of replenishment quantities, increase the efficiency of replenishment processes, and ultimately optimize inventory turnover rate by using technologies such as big data integration and analysis, advanced predictive models and replenishment optimization algorithms, and automated process flow.

[0004] The following is the specific plan.

[0005] like Figure 1 As shown, the main roles and corresponding business processes are as follows.

[0006] System: Processed by the system (non-human role), sales forecasting is performed according to a predetermined model, replenishment suggestions are calculated using a predetermined optimization algorithm, and replenishment orders are automatically generated through a predetermined process, containing the recommended replenishment items and quantities.

[0007] Store level: This is handled by the store manager, who usually reviews the replenishment order generated by the system for the day, modifies and confirms the replenishment items and quantities suggested by the system, and displays the calculation basis such as future sales forecasts, current inventory in stock and in transit, and various influencing factors for reference. After final modification and confirmation, the replenishment order can be submitted for confirmation.

[0008] Supply Chain: This is handled by the supply chain manager, who typically reviews replenishment orders submitted by stores, approves the replenishment items and quantities, and displays the calculation basis (as seen in the store) and the available inventory in the store's corresponding warehouse for reference. After final approval, the replenishment order can be confirmed.

[0009] After that, they are transferred to various warehouses for shipment according to replenishment orders. Attached Figure Description

[0010] Figure 1 Roles and business processes.

[0011] Figure 2 Store replenishment suggestion process and calculation method.

[0012] Figure 3 Store sales forecasting (forecasting model and selection).

[0013] Figure 4 Product lifecycle and weekly sales volume. Detailed Implementation

[0014] To better describe the intelligent replenishment algorithm and process design involved in this document, it will be described in sections below. Figure 1 Each business process within it.

[0015] 1. Sales Forecast like Figure 3 As shown, based on historical sales data, product data, and influencing factors, various prediction models are substituted to predict sales for the next four weeks (including this week). Then, the prediction results of various models are compared and analyzed with the actual sales based on the measurement objectives to find the optimal model applicable to each product life cycle.

[0016] in, 1) Historical sales data will be collected from the sales system, including total sales and average daily sales over the past four weeks.

[0017] 2) Product data, including product lifecycle and product sales tiers, as detailed below.

[0018] Product Lifecycle Refreshed weekly, divided into: New product launch period: Similar to the previous generation, it is configured according to the product series and is recorded as week X.

[0019] Stable period: After the new product label is maintained for X weeks (i.e., after the new product period), it enters the withdrawal period.

[0020] Exit period: Rolling 2-week average < Historical average sales AND This week's sales < Historical average sales AND This week's sales < 1 / 3 of the maximum rolling 2-week sales.

[0021] Basic as follows Figure 4 The weekly sales figures are used to divide the period.

[0022] Product Sales Tiers Product sales tiers: Updated weekly, categorized based on sales over the past four weeks for periods of stability: Category A: Best-selling products, accounting for the top 60% of sales (SKU count accounts for approximately 10-20%).

[0023] Category B: Mid-range products, accounting for 30% of sales volume (SKU quantity accounts for approximately 20-30%).

[0024] Category C: Slow-moving products, accounting for 10% of sales volume (SKU count accounts for approximately 50-70%).

[0025] 3) Influencing factors, including promotion factors, holiday factors, and customer flow factors.

[0026] 4) Predictive models, including exponential smoothing model, Arima model, Holtwinters model, Prophet model, random forest (RF), GBDT, Xgboost, and Deep Learning.

[0027] 5) The evaluation target is to select the prediction model with the highest accuracy as the optimal model for the product lifecycle.

[0028] Prediction accuracy: Class A 70%, Class B 60%, Class C 50% Calculation method: Prediction accuracy = Number of SKUs with a deviation rate <= 20% / Total number of SKUs in this category, where a deviation rate of <= 20% for a single SKU is expressed as: (Estimated sales - Actual sales) / Actual sales <= 20%. 6) Model adoption: On the one hand, the optimal model for each product life cycle is evaluated, and on the other hand, the predicted sales of the optimal model are used as the final future sales forecast result.

[0029] 2. Replenishment suggestion calculation like Figure 2 As shown, through various inputs and processing, the replenishment items and suggested replenishment quantities are calculated.

[0030] 1) Input Basic store data, such as historical sales, available inventory (in stock - locked), shipped but not yet received (in transit), ordered but not yet shipped (in transit), transferred out in transit (out), transferred in transit (in transit), target DOS, and highest and lowest water levels, are used as inputs.

[0031] 2) Processing Based on the above inputs, including historical sales, system configuration factors (such as promotions, holidays, and customer traffic), product lifecycle, and product sales tiers, the estimated sales for the next four weeks are calculated using the optimal model evaluated based on the current stage of the product lifecycle.

[0032] 3) Calculation method for replenishment suggestions Based on the above inputs, including available inventory, in-transit inbound, in-transit outbound, target DOS, highest water level, lowest water level, and projected sales for the next 4 weeks, calculate the recommended items and quantities for store replenishment.

[0033] 4) Output Generate the recommended replenishment items and recommended replenishment quantities for the day.

[0034] 3. Generate a replenishment order (suggested replenishment items and suggested replenishment quantity). Based on the suggested replenishment items and quantities generated above, a replenishment order for the day is automatically generated, which stores can modify, confirm, and submit.

[0035] 4. Confirm the replenishment order (confirm the replenishment items and quantity). Based on the daily replenishment order generated above, which includes suggested replenishment items and quantities, the store (usually the store manager) modifies, confirms, and submits it. If it is not confirmed and submitted within the deadline, the system will automatically confirm and submit it.

[0036] 5. Approve replenishment orders (approve replenishment items and replenishment quantities) Based on the aforementioned daily replenishment order, which includes confirmed replenishment items and quantities, the supply chain (usually the merchandise manager) will approve it. If it is not approved within the specified time, the system will automatically process the approval.

[0037] 6. The warehouse issues goods according to the replenishment order (approved replenishment items and quantity). After that, they are transferred to various warehouses for shipment according to replenishment orders.

Claims

1. A smart replenishment algorithm and process design for retail stores. This invention provides an optimized algorithm and processing flow for smart replenishment in retail stores, involving functions such as store sales forecasting, store replenishment suggestion calculation, and smart replenishment order generation and circulation.

2. The intelligent replenishment algorithm and process design for retail stores according to claim 1, through in-stock inventory model, sales forecasting model, multi-factor fusion and superposition, and suggested replenishment optimization algorithm, predicts future sales, suggests replenishment quantity, and automatically generates replenishment orders, providing effective suggested values ​​for store replenishment, improving replenishment efficiency, and optimizing inventory turnover.

3. The intelligent replenishment algorithm and process design for retail stores according to claim 1, by introducing advanced models, optimization algorithms and automated process technologies, can significantly improve the accuracy of replenishment quantity and the efficiency of replenishment process, and ultimately optimize inventory turnover rate.

Images