Store-Level Replenishment and Inventory Planning in Retail

In brief: A full warehouse and an empty shelf is a situation frequently encountered in retail. In retail, inventory turns into sales not in the central warehouse but on the shelves of hundreds of stores. Demand at the store level is low and volatile, shelf space is limited, and replenishment depends on pack size and the shipping calendar. This article covers calculating the store replenishment quantity, the role of display stock, distributing stock between stores and the warehouse, and how initial allocation differs from replenishment.

How Is the Store Replenishment Quantity Calculated?

The most commonly used method in store replenishment is the periodic review "order-up-to" policy. In this policy, order quantity = target level − (shelf stock + on-order). The target level is determined jointly by demand over the protection interval, safety stock, and display stock. An example is given in the table below (figures are for illustration purposes):

If the target level falls below the display stock (40), the display stock is used as the target. The reason is that the minimum quantity needed for the shelf to look full is a constraint independent of the sales forecast. In addition, for slow-moving products with a large case pack quantity, pack rounding creates permanent excess stock in the store. For these products, evaluating the options of breaking cases or less frequent shipments is recommended.

Why Should Display Stock Be Handled Separately?

• It is unrelated to sales: It is the minimum quantity that must be displayed on the shelf; for low-selling products it can make up the largest part of the stock.
• It should not be confused with safety stock: Display stock is available for sale, but the aim is not to go below that level so that the product does not "look empty".
• It is determined together with the planogram: As the number and depth of shelf facings change, replenishment parameters must also be updated.

Should Stock Be Held in the Store or in the Warehouse?

As demand is pooled, uncertainty decreases. A product that sells less than expected in one store may sell more than expected in another; therefore, the sum of fluctuations in individual stores is greater than the fluctuation seen at the central warehouse. When the portion of safety stock that can be held is kept in the central warehouse and stores are replenished frequently in small quantities, total inventory decreases. The limit of this approach is set by logistics cost, the store's receiving capacity, and the need for display stock.

Every unit held in a store serves only that store's customers for as long as it is unsold;
a unit held in the warehouse serves all stores.

The Difference Between Initial Allocation and Replenishment

There is no past sales data for fashion, seasonal, and campaign products. For this reason, initial allocation is made according to store cluster, store size, and the performance of similar products; part of the stock is deliberately held back in the warehouse. When data from the first weeks of sales becomes available, a second allocation is made to well-selling stores. For basic (continuous) products, the replenishment logic above applies. One of the mistakes frequently encountered in practice is managing these two product groups with the same parameters.

Which Metrics Should Be Monitored?

Technology and Digitalization: Automatic Replenishment, RFID, and Computer Vision

Store-level replenishment requires decisions to be made every day for thousands of product-store combinations. Since it is not possible to make most decisions manually at this scale, automatic replenishment takes on this task. However, the success of automatic replenishment is limited by the accuracy of the inventory data and parameters it is fed with.

Enterprise Resource Planning (ERP) is the enterprise transaction system that records orders, inventory, purchasing, production, and finance transactions in a common database; it is the master data and transaction source for planning. It holds point-of-sale (POS) data, product master data, and order records, and forms the basis of automatic replenishment calculations.

Warehouse Management System (WMS) is the system that manages goods receipt, putaway, picking, and shipping in the warehouse on a location basis. It manages picking and shipping in the warehouse according to store orders and ensures correct preparation at the case and pallet level.

Radio Frequency Identification (RFID) is a tagging technology that allows items, cartons, or pallets to be read in bulk by radio waves without requiring line of sight. It speeds up store stock counting and improves record accuracy; thus, orders that are not placed because of incorrect inventory records are reduced.

Internet of Things (IoT) is the infrastructure in which sensors on machines, vehicles, warehouses, and products transmit data automatically over a network; location, temperature, status, and consumption data are collected without human input. Shelf status and product freshness are monitored through cold chain temperature sensors and smart shelves.

Artificial Intelligence (AI) and Machine Learning (ML) are methods that learn patterns from data to produce forecasts, classifications, and recommendations; machine learning is the sub-field of artificial intelligence that learns from data. They produce forecasts at the product-store level and include the effects of weather, local events, and campaigns in the forecast.

Computer Vision is an artificial intelligence application that extracts information such as shelf gaps, planogram compliance, or product counts from camera images. By detecting shelf gaps and planogram non-compliance, it makes "in the system, not on the shelf" situations visible.

Transportation Management System (TMS) is a system that supports shipment planning, vehicle loading, routing, and carrier management. It plans the routes and vehicle fill of store shipments.

In trainings, it is often asked what the planner's role will be in an environment where orders are placed by the system. Automatic replenishment supported by artificial intelligence does not eliminate the planner; it changes the planner's job description. Routine orders are placed by the system, while the planner manages parameters, exceptions, and new product and campaign situations. For this transition to be made soundly, it is recommended to measure store inventory accuracy and to monitor in which situations system recommendations are manually corrected.

Automatic replenishment relies on the stock in the system, while the customer relies on the stock on the shelf;
when the two do not match, the order is not correct either.

Implementation: Checklist

1. Separate products into basic (continuous) and seasonal/campaign; manage them with different processes.
2. Set store replenishment parameters (review, lead time, service target) by product segment.
3. Take display stock from the planogram and add it to the replenishment calculation as a separate constraint.
4. Measure the effect of case pack rounding on slow-moving products.
5. Evaluate the distribution of safety stock between stores and the warehouse together with logistics cost.
6. Measure on-shelf availability and lost sales regularly.
7. Measure store inventory record accuracy regularly by counting or RFID; the correct operation of automatic replenishment depends on accurate inventory data.

Related Articles

• 5 Common Mistakes in Safety Stock Calculation and the Correct Approach
• Demand Planning and S&OP in Fast-Moving Consumer Goods (FMCG): 5 Critical Topics
• How to Forecast Promotion Impact: Campaign Types and Uplift Calculation

Frequently Asked Questions

Should forecasts be made at the store level or at the warehouse level?

Since the replenishment decision is made at the store level, forecasts are also needed at the store-product level. However, data at this level is quite volatile. The practical solution is to build the forecast at a higher level (product group, region) and allocate it to stores proportionally.

Does display stock replace safety stock?

It does so partially. For slow-selling products, display stock is often larger than safety stock and effectively provides protection. For fast-selling products, safety stock is added on top of display stock.

How are lost sales estimated?

First, the days on which the product was not on the shelf are identified; then the expected sales on those days are calculated from the sales rate on in-stock days. This estimate should also be used to correct the training data of the forecasting model; otherwise, the model interprets stockouts as low demand.

How often should stores be replenished?

Replenishment frequency depends on sales rate, shelf capacity, and logistics cost. In general, frequent and small replenishment for fast-moving products and less frequent replenishment for slow-moving products reduces total cost.

Do automatic replenishment systems eliminate the need for a planner?

Automatic replenishment systems do not eliminate the need for a planner. The system takes on routine orders; parameters, exceptions, and new product and campaign situations remain the planner's responsibility. In this structure, the planner's role shifts from doing calculations to managing the process.

References

• Fisher, M., & Raman, A. (2010). The New Science of Retailing. Harvard Business Press.
• Silver, E. A., Pyke, D. F., & Thomas, D. J. (2017). Inventory and Production Management in Supply Chains (4th ed.). CRC Press.
• Eppen, G. D. (1979). Effects of centralization on expected costs in a multi-location newsboy problem. Management Science, 25(5), 498-501.

Related Optimum Planlama Trainings

Institutions We Work With in Retail

Some of the institutions to which we provide training and/or consulting services in the retail and store operations sector are listed below. For all sectors, you can visit our References page.

BR Mağazacılık, 2014, 2015

Civil Mağazacılık, 2019

LCWaikiki, 2010–2014

Makbul Kuruyemiş, 2021, 2022

Migros, 2021

Paşabahçe, 2014–2016

Zen Pırlanta, 2022

Comments

“It was very useful to apply many of the prediction models that remained in the theory and to use them by assimilating their content. However, the explanation of theoretical issues based on clear foundations was very successful. Talking about current examples parallel to the functioning of the institution/sector, making analyzes on them was effective in forming different perspectives and ideas.”

E**** K****, Supply Chain Planning - Manager, Migros (Demand Planning, 2021)

“Mr. İlker, Thank you very much for the training. It was really a very instructive and productive training. I came across many things that I was missing in what you told me and that I applied some of them without knowing them by name. I will look at my notes again and again in order not to forget them in the process. We hope our paths cross again in the future.”

Aykut Köroğlu, Supply Chain Planning - Department Supervisor, Migros (Demand Planning, 2021)

“I believe that Mr. İlker used time, shared his experiences with us, enjoyed receiving training from a special consultant, and that this training was very useful for me as a colleague and added experience to me. I would like to express my endless thanks.”

Melike Orhan, Supply Chain Planning - Expert, Migros (Demand Planning, 2021)

To view all testimonials, click here.

Request a quote for training or consulting tailored to your organization.