Automatic stock replenishment

Store stock replenishment is important and how the replenishment is done impacts the sales directly.

There were a lot of researches carried out on impact of not-on-shelf, out-of-stock(OOS) products on the retail companies' sales on different types of products. The researches were carried out on replenishment of products from DC, from back-rooms of the stores, replenishment modules and most important how customers react when they do not find the product they need.

The reaction of customers (on OOS) differs a great deal. If the customer buys a different brand, we are happy. If she or he does not buy anything at all, then we are not content. And if the customer buys the product on the competitor's store, that is catastrophe! (The Impact of Automatic Store Replenishment on Retail, Alfred Angerer, page 4)

It was calculated for German grocery retail store that around 30% of customers purchase items, they did not find, from another competitor's stores. The most important the frequency of OOS products. In fashion retail, the right-time, the right-quantity is more important. The merchandise sales depend on weather, people mood, trend, day of the week, time of the day, etc, too many variables in the formula. Some general replenishment rules, however, can be applied for never-out-of-stock(NOS) products, like socks, men underwear etc.

The formula to calculate the store stock level is:

E=D+CR-RTM-CW-S&D-S-RTW-TRSF

Where:

D: Delivery to the stores

E: current stock level at the store

CR: Customer returns

RTM:Return to manufacturer

CW: Waste

S&D: Soiled and damaged products

S: Sales

RTW: Returns to Warehouse

TRSF: Transfers between stores

To make the order replenishment: D=CS-E

where CS is planned quantity for the stock level in the store and D the quantity to be delivered to the store. This is the most basic formula. In most retail companies: variables derived from store RPM, store last season sales, promotional activity, distance from the DC(replenishment rate per specific store) is added to the formula making the OOS frequency most minimal.

This formula can be applied on SKU level or can be used as summary. The most important is how this information is entered. Suppose warehouse has created an order for delivery to stores on Thursday according to Monday-Tuesday stock movements. The store stock movements for Wednesday-Thursday are not updated on the company system yet. So the order created does not consist of the merchandise sold on Wednesday-Thursday, two day delay!

CS on a unit base for models consisting of size can be calculated automatically by analyzing the last-season-sales or can be entered by merchandisers manually for a product group. For example: A model consists of 5 sizes: S, M, L, XL, XXL. The merchandiser can let the system calculate the size distribution by the last season sales of the product group the model belongs to or can enter manually entering the percentage per model. For example: S-15%, M-25, L-25%, XL-20%, XXL-15%. After the target is set the only thing is to set how many items can be displayed in the sales floor of the store. The system will distribute the sizes according to the percentages set or calculated by the system. The formula can be extended by entering more variables. Suppose a store is located 1500km away from DC and the RPM rate is considerably high. The merchandiser enters only how many items of a particular model can be displayed at a time so the above formula will not work for a store located 1500km away because there will be only few deliveries in the week. We need to add also a store back-room stock so store personnel can replenish sales from the store's back-room. We need to add an RPM variable to the formula also. Suppose there is a store 1500km away but its RPM is three times lower than the average stores RPM. Do we need a back-room stock for that store? But what if the RPM of the store is 4 times higher than the stores's average RPM and it is 1500km away from DC? The stock collected at WH should be divided into two types of containers: ones for immediate display in sales floor, the other ones are for sales replenishment from the back-room. The containers should have unique numbers on them and every hour stores personnel should get a picking list automatically generated from back-office products to replenish the sales. But if the store is near to DC(50-75km range) no need to add distance and RPM variables to the formula.

The above table is the replenishment ratio per a year. Unlike fmcg goods, it is not important to have 100% replenishment on sales months. The average replenishment ratio is 100% per a year.

In the above graphic, the stock in the store is divided by season. In the beginning of the summer season the stock for summer season merchandise is maximum but as we get closer to autumn the stock for summer merchandise decreases but for autumn merchandise increases. If we dig deeper into the sales/replenishment ratio the replenishment ratio decreases for summer merchandise as we got closer to autumn months. The catastrophic situation if at the beginning of summer season the replenishment ratio for summer season merchandise is low. For NOS products no seasonal approach is not needed.

Across-the-dock concept

Across-the-dock method is a kind of order-fulfillment operation in which some time-consuming warehouse operations are omitted.

The objective for order-fulfillment operation is:

The order fulfillment operation objective is to ensure that SKUs meet company quality standards, which require that the correct SKU is transferred from the storage position to the correct pick position, in sufficient quality and at the appropriate time. (Order-Fulfillment and Across-the-Dock Concepts, Design, and Operations Handbook, David E. Mulcahy and John Diletz, page 9)

To summarize the order-fulfillment operation objective:

• withdrawn in the right quantity
• in the correct condition
• on schedule
• must have a packing list
• be packaged in a protective and labeled shipping container
• be properly manifested
• be delivered to a customer delivery location within the customer-order and delivery cycle time

Failure in one of the above requirements makes the full process unsuccessful: more or less items delivered to the customer, damaged goods were delivered, early or late delivery, goods were sent to another location, etc.

The most important issue is how much money is spent on the above tasks. If all the tasks accomplished according to company standards but the cost of operation is high, again, the full operation is unsuccessful. Because order picking cost consumes most of the warehousing budget and in some countries is the first or second highest expenditure after the cost of rent of warehouse itself. Suppose company sells LCD TVs for 1000 USD and cost of picking and delivering it to the customer is 10 USD. The cost of order-fulfillment is 1% of the sales so it is quite acceptable. But if company sells lingerie items average for 10 USD per item and order-fulfillment cost is 1 USD per item, which is 10% from sales. So for the order-fulfillment definition we have to add a phrase: and at the acceptable cost.

We have an objective for order-fulfillment and explained why this operation is important in logistics and company business. Some tasks of Order-fulfillment can be done at supplier site, in warehouse while accepting the delivery from supplier or after placing it to the storage area. The order-fulfillment done while accepting the delivery from supplier can be, in some cases, a cost-effective operation. Suppose jeans wholesaler has a truck of jeans of inbound. The inbound operation takes two days, picking of 15-25 orders takes another 2-3 days and transportation of merchandise takes another 1-2 days. The time from accepting inbound truck to delivering to the end customer is 5 to 7 days. In the end 80% of the truck is sent to the customers. The truck consists of 30-40 different jeans models.

Suppose the same operation is done while inbound operation: jeans scanned into piles and scanned piles sorted according to pre-orders from customers and the rest 20% is put to the storage area in no more than four days. At the end of inbound the orders are collected and scanned. By implementing sorting we can omit picking and reduce storage area significantly. This method is called across-the-dock method.

Suppose 10 jeans are scanned into pile and size distribution is as bellow:

1. 32-30
2. 32-30
3. 32-30
4. 32-30
5. 32-32
6. 32-32
7. 32-32
8. 32-34
9. 32-34
10. 32-34

The size 32-34 is scanned the first in the pile so it is in the bottom of the pile and the size 32-30 is scanned the last so it is in the top of the pile. The pile has a unique number (permanent address) and when operator who scans the pile starts a new pile by entering a new number for the next pile the system generates a sorting list for the pile just scanned and prints it.

The sorting list will be as bellow:

1. 32-30 ---- 1
2. 32-30 ----2
3. 32-30 ----3
4. 32-30 ----4
5. 32-32 ----1
6. 32-32 ----2
7. 32-32 ----No
8. 32-34 ----1
9. 32-34 ----2
10. 32-34 ----3

------------------1 -----------------

The second column is the conveyor or container place numbers where these sizes are to be distributed. At the end of the list in bold is the quantity of jeans which are not distributed so if at the end of sorting the employee counts less or more than this quantity then he/she made a mistake while distributing the piles or the there was an error while it had been scanned. The flow work correcting/controlling the mistakes will be discussed later. The rest of the pile is stored in a separate location. Also there is no need to print the outbound order number. The only thing employee needs to know it what item to put at which location. The system will link an order number to the location number. The order number will be printed on the packing list while it is scanned to the transportation containers. The sorter can be manual, semi-automated, and fully-automated. In our case, the sorting done manually because start-up costs are low and the employee salaries are low but in some countries fully-automated sorter would be more reasonable than hiring employees for sorting.

The sorter type can change from goods type. In our case, it is a vertical shelf like structure of 1.5m depth having an opening on the other side and a decline of 15 degrees with gravitational rolls on the bottom. The goods are scanned to the transportation containers from the other side of this structure. More simple horizontal sorter can be used. Goods can be sorted on separate pallets or containers but this time scanning and marshaling will be difficult and this system is area consuming system but cheap one.

Wholeseler's warehouse operations vs retailer's

A few weeks ago I had a chance to analyze jeans wholesaler’s warehouse operations like: inbound and outbound. A wholesaler has generally two main peaks: before spring-summer season and before autumn-winter season. Of course there are other peak periods but compared to these two main peaks they are small ones. In these peaks warehouse area increases a lot and man work power is needed much more than ordinary work days. Because of this wholesaler rented a warehouse which 2/3 of the year was almost empty and wholesaler employed much more warehouse employees to cover these two main peaks in his business. In addition to wholesale operations wholesaler has his own operated retail stores selling jeans under his own brand in different shopping malls and high street stores around the country. So a small part of the stock is kept in warehouse to feed these stores and for re-orders from wholesale buyers.

Because there are about 150-200 different models per season, there is no big need for WMS system in warehouse. There are barcode labels on products and inbound operation was done by scanning the products, making piles of 10-15 pieces of one model and putting them to the storage area right after the scanning. There is no addressing system so rows in the storage area are divided like man jeans area, ladies jeans area, ladies tops, etc. Every model’s all sizes are kept near to each other and are not mixed with other models unless the quantity is too small and there is no storage area for new delivery from supplier. Scanning is done just to know what received, but not to know where in the warehouse this product is kept.

The size range for jeans trousers products is too big compared to tops. Every jeans trouser has 20 different widths. Every width has 5 different lengths. So per one model there can be 100 different sizes.



Suppose warehouse has received 30 pieces of jeans of one particular model. They were scanned into five piles, every pile consisting of 6 jeans. Suppose 24-30, 27-32, 27-34, 33-36 are to be collected from these 5 piles. Finding these 5 piles among other 200 models would be a little big time consuming. Finding a particular size among these five piles will consume another time. 24-30 is on the third pile, 27-32 is in the 1st pile, 27-34 in on the last pile, 33-36 is one the 1st pile. On the 2nd and 4th pile there are no products to collect. The picker generally searches all 5 piles in order to find these 4 jeans, although there is no need to search 2nd and 4th piles.



Suppose there is an address for every pile. For example, these five piles are located in 101 numbered cell and 101 numbered cell is divided into five piles and each pile has a unique one digit letter within this cell. The picking list would be like this:

101-----a------33-36----- 1 piece
----------------27-32----- 1 piece
---------c------24-30----- 1 piece
---------d------27-34----- 1 piece

So the picker omits searching b(2nd pile) and d(4th pile).


If we somehow able to index the items in the pile while scanning them on inbound, we can increase the productivity of pickers a bit more by showing the exact location in the pile from the top. The picking list would be as following:

101---------a-------4------33-36--------1 piece
---------------------5------27-32--------1 piece
-------------c-------4------24-30--------1 piece
-------------d-------6------27-34--------1 piece

The picker counts 3 jeans from the top, takes the 4th one, checks if the right size. Then, takes the fifth jeans. So he omits the checking of the first 3 top jeans in the pile (a). Next, counts first 3 jeans from the top in the pile (c), takes the forth one, check if the right size and so on.

Suppose there is second order to collect(blue items). Suppose first order(red ones) is collected and after this collection the piles will look like below:


The picking list would look like this:

101-------a-------1--------24-34-----------1 piece
-------------------4--------27-32-----------1 piece
-----------b-------4--------36-36-----------1 piece
-----------d-------4--------30-36-----------1 piece

There is another way to collect these two orders. These two orders can be collected at the same time.

The picking list for both orders combined would be similar to this:

Cell number		Pile letter	Index		Size	1st order	2nd order
101		a	1		24-34		1 piece
			4		33-36	1 piece
			5		37-32	1 piece	1 piece
		b	4		36-36		1 piece
		c	4		24-30	1 piece
	d		4	30-36			1 piece
	e		6	27-34		1 piece

In this case the picker sorts these two orders at separate containers while picking.

The picking can be combined and order sorting can be done manually while scanning or an automated sorter can be used. The picking list would be as following:

Cell number		Pile letter	Index		Size	Quantity
101		a	1		24-34	1 piece
			4		33-36	1 piece
			5		27-32	2 pieces
		b	4		36-36	1 piece
		c	4		24-30	1 piece
	d		4	30-36		1 piece
	e		6	27-34		1 piece

The picking can be combined in a way so sorting at the end of picking can be omitted:

Cell number	Pile letter	Index	Size	Order number	Quantity
101	a	1	24-34	2nd	1 piece
		4	33-36	1st	1 piece
		5	27-32	1st	1 piece
		6	27-32	2nd	1 piece
	b	4	36-36	2nd	1 piece
	c	4	24-30	1st	1 piece
	d	4	30-36	2nd	1 piece
	e	6	27-34	1st	1 piece

The collected pile would be like this:

27-34
30-36
24-30
36-36
27-32
27-32
33-36
24-34

In the bottom of the pile will be located items from the top of the picking list. If the pile of goods collected is unstable the picker may start another pile but put a mark in picking list where the new pile begins from. After he collects all the goods in the picking list, he sorts the goods according to orders containers starting from the last collected pile. A separate container for every order in the picking list is created. This method is good for order picking between 10-30 orders at once.
In fact, there is another method of accomplishing outbound order without picking process and this method is called across-the-dock method (not a cross-dock).