Theory and Practice of Supermarkets – Part 2

grocery store market supermarket retail shop
How to use supermarkets correctly …

In my last post, I described how supermarkets work in theory. But while knowing the theory helps, actually creating a working supermarket is much more difficult. Are there situations where supermarkets are not so useful? (Hint: Yes!). And what is needed to have a working supermarket? Let’s find out!

What Are the Advantages of a Supermarket?

Supermarkets have a number of advantages compared to “normal” inventory. Regarding pull systems, they are designed to give a signal to production if material leaves the system. However, this can also be implemented in other ways.

If your supermarket is implemented physically and not only through digital sorting in an ERP system, you can get an overview of your system through a quick glance at the supermarket. It is usually easy to see which inventories are high, low, or dangerously low. Hence, a supermarket is often very helpful for visual management.

Probably the biggest advantage of a supermarket is its FiFo principle. By maintaining FiFo, the oldest parts are sold first. Thus you are less likely to run into problems due to old parts being left over somewhere in the inventory.  Additionally, if you have a quality problem, it is often easier to track if the parts come down the system in a FiFo fashion.

Supermarkets in Practice FAQ

How big should the supermarket be?

The question of the size of the supermarket is a tricky one. Ideally it should be able to fit the material on all kanban cards (i.e., they should be able to hold all of the WIP that is permissible in the system). Hence your supermarket is big enough to fit all your kanban cards. As for the number of kanbans, this is actually a bit more tricky. See my posts on the kanban formula, or the easier approach with kanban estimation.

However, in a normal system, many of the kanban cards would be waiting for processing or currently processed. It is possible but unlikely that your supermarket is completely full. Hence, reserving all the space for inventory that you rarely need may be wasteful. I also have seen supermarkets that were too small for all the inventory but could hold the current inventory most of the time. In the rare cases of a full supermarket, the material was stored in a general area as a sort of overflow storage.

Overall, both are possible. You can size the supermarket to the maximum WIP, or lower. The first one takes up more space. The latter takes up more effort to handle material and maintain FiFo in case of overflow. It depends on your circumstances what is best. My preference, however, would be to improve the system so you can work with less WIP altogether.  🙂

Where should the supermarket be located?

manager and worker
Let’s put it right next to where I make the parts …

The supermarket is managed by the processes that supply the supermarket. To reduce the distance for both the material and the information flow, supermarkets are best located at the end of the supplying process. If you locate the supermarket at the location of the customer, it will be more cumbersome. Hence, supermarkets should be, whenever possible, close to the last process on the supply side.

How about multiple lanes for the same parts?

To maintain FiFo by hand, some handling is needed. Many supermarkets make this easier by using rolling lanes or slides. Parts added at one side roll or slide down the inclined lane toward the other end. Such rolling systems can be used for many different-sized parts, from small plastic boxes to full-sized pallets.

For pallets and items in rolling containers, I have also seen a sort of railings, where the carts or pallets are pushed in on one side. With pallets, the forklift drivers usually push them in using forklifts. Due to the railings, they cannot leave the track but maintain FiFo.

Upside Down Flower Pot
Here, there, elsewhere …

A small problem is if there is not enough space in one of these lanes for all the parts of this type. Of course you could reduce the inventory through SMED or other improvement methods. Nevertheless, such multiple parallel lanes with the same part type are found frequently in industry. The problem with these is maintaining FiFo. I have seen a number of solutions, including simple upside down plastic flower pots in different colors on the lanes that are to be used, mechanical barriers that allow removal or adding only in one lane, or even fancy digital systems.

Forklift in Operation
Keep your flower pots away from me!

I am usually not convinced. Even with such a system, the FiFo can be mixed up if there is very little or very much material. Most of these systems are also an additional burden to shop floor logistics. Especially the upside-down flower pots are a hassle. For example, forklift drivers would have to get out of their vehicle to move these pots.

Believe me, you will not be popular with the forklift drivers. I would be willing to push for it if there is a clear advantage by maintaining FiFo. This could be, for example, with quickly aging products (milk?), frequently updated products, or if FiFo really helps you track problems in your particular case.

However, I have in the past also opted to break FiFo and allowed both adding and removing randomly in whatever lane the operator chooses. The random adding and removing is not quite FiFo, but in these cases there was no system in place to track problems anyway (which was another problem). Additionally, the products did not really age quickly, nor was their design changed frequently. Although if there was a design change, special care had to be taken so as not to mix up the old and the new products. I agree, it is not a perfect solution, but at that time it was in my view the best option.

When to use a supermarket, when a FiFo?

Kanban Loop options for three Processes
Kanban loop options for three processes

When transforming a value stream into a pull system, there is always the option between two processes to split a loop in two by adding a supermarket or to keep one loop by adding a FiFo lane. I did some original research on this. For details, read my posts on Ten Rules When to Use a FiFo, When a Supermarket – Introduction and the ten rules.

The rules in brief are: Use a FiFo whenever there is no reason to use a supermarket. The ten possible reasons to use a supermarket are:

  1. Supermarket for process-specific lot size differences
  2. Supermarket in front of the customer
  3. Supermarket if material flow splits up into different directions
  4. Supermarket between very different cycle times
  5. Supermarket between different shift patterns
  6. Supermarket when creating different variants
  7. Supermarket for merging of material flows
  8. Supermarket for large distance between processes
  9. Supermarket in the case of high demands on flexibility and reaction time
  10. Supermarket for change of responsibility

When is a supermarket NOT useful?

Different Screws
Low quantity, high variety
Identical Screws
High quantity, low variety

Supermarkets can help you in managing your inventory at the end of the pull system. However, they also have limitations. Supermarkets are well suited to high-quantity, low-variety products. Since they are used in combination with kanban, there should be a continuous demand for more of the same product. A lane in the supermarket should have only one part type, and that part type is always reproduced after consumption.

However, not all products are high quantity, low variety. Many industries have quite the opposite, a low-quantity, high-variety production. In the extreme, every part may be unique for one individual customer, as, for example, many machine tool makers. In this case it makes no sense to assign an entire lane in the supermarket to this single product, which is produced only once in a lifetime. In fact, the entire kanban system is the wrong approach to these type of products.

Roller coaster ride
CONWIP -go-round with different parts – like roller coaster seats with people

Luckily, in this case, CONWIP systems can be used. CONWIP works very similarly to kanban, but CONWIP cards are not assigned to individual products. Rather, they generically permit the production of the next most urgent job in the queue.

Luckily, it is pretty easy to combine hybrid kanban-CONWIP systems by having both types of cards. The kanban cards would go in a supermarket, and the CONWIP cards would go in an ordinary type of inventory.

But do not use supermarkets for high-variety, low-volume products!


Initially I wanted to write a brief post on supermarkets, hoping that I get enough material to cover around 1000 words. As it turns out, there is a lot more to supermarkets! These two posts together have more than 2500 words, even with only linking to some more detailed topics like the number of kanban or the decision where to place supermarkets. If you liked this article, you can also check my related posts on Theory and Practice on FiFo Lanes and the Top Five Cases When NOT to Use a FiFo. In any case, I hope this helps you to go out and organize your industry!

One thought on “Theory and Practice of Supermarkets – Part 2

Add a Comment