SBP 10a: Inadvertent JIT

This principle occurs because customers are aware of how long their inputs have been in the production process, have expectations for how long production should take, and are intolerant of avoidable delays.

We usually think of inventory as stored goods. Why do we keep inventories? There are a number of reasons, including

• production smoothing (meeting varying demand with stable production, which is helpful even if the demand can be perfectly predicted or scheduled),
• process decoupling (creating an inventory buffer between stages of a production process),
• realizing economies of scale (such as order quantity or production batch size advantages),
• maintaining buffer stocks (extra product to be used in the case of higher than normal demand), and,
• simplifying material handling (it might be very difficult to move production in quantities of one).

All of these reasons are included in the idea that inventories are used move items forward in time from the point at which they are available until the point at which they are needed. Inventory is a temporal transformation process, or in other words, inventory moves things forward through time. Why do we need to move things forward in time? Because we are unable or unwilling to properly match production with demand.

Manufacturing companies that are willing and able to match production with demand do so by implementing just-in-time (JIT) production systems. With JIT systems, production is scheduled and controlled so that it occurs just prior to the time it is needed. Manufacturers often implement JIT systems by utilizing a signaling device known as a kanban or card. The “downstream” portion of the process sends a kanban card to the “upstream” supplier requesting more production. The upstream process produces the amount requested on the card, sends the card and the product to the requester, then waits for the next card (or does something else for the time being). The kanban card is therefore a signal for a production station to produce. The result is dramatically reduced inventories, since the upstream supplier only produces what is needed. In other words, effective JIT systems avoid the need for the temporal transformation that comes from production which is not in sync with demand.

Warehousing and storage services exist to temporally transform customers' belongings-making them available at a later date. Also, a primary reason retailers and wholesalers exist is to stock manufactured goods until the customer wants them. A retailer might have an item in inventory for months before it is sold. However, when a customer visits the retailer and selects the item, that customer does not want to be delayed in a queue for an extended period of time. This disdain of delays tends to be true of all services in which customer inputs enter a queue.

When, then, is a queue? A queue is a waiting line. It can be a waiting line of people, such as the line waiting to buy tickets at the movie theater; it can be a waiting line of things, such as the rolls of film waiting to be processed at the photo processing lab; or, it can be a waiting line of information, such as the line stack of loan applications waiting to be processed or the buffer of Internet queries waiting to be transferred (which is why the Internet can be slow during some times of the day.) In all of these cases, queues exist because demand has temporarily exceeded the rate of production. Sound familiar? A queue is in fact a short-term inventory system! (This is why the Customers in Inventory Service Business Principle indicated that services can be inventoried–the inventories are manifest as queues!)

There are, in fact, some fundamental differences between queues and traditional conceptualizations of inventories. In the business sense queues are typically described as waiting which belongs to a customer, whereas inventories simply belong to the company (until sold). This we see in the example above, with queues of customer-self inputs, of customer-belonging inputs, and customer-information inputs.

An implication of customer ownership of queue items is the customer concern about waiting times. For manufacturing, inventories are evaluated in terms of “inventory turnover,” which is the number of times that inventory is replace in a given period. This is a function of the amount of time items spend in inventory. It is not unreasonable for an average manufactured item to spend weeks or months in inventory. Not so for customer-input inventories. Customers are often enraged if we make them or their belongings wait weeks or months in a queue! (Imagine spending a week waiting to be served at the bank or a month waiting to have your home's air conditioner repaired!)

The result of customer impatience is that we often measure queue performance in terms of seconds or minutes waiting. (Although sometimes hours or days.) In other words, we require the production system to be no more than a small amount of time out of sync with demand–which is the essence of a JIT system! This JIT system comes not as a result of management initiative, but rather is motivated by market necessities (customers would depart if they had to wait more than a short time for service). Therefore the JIT approach taken by service companies is to a large extent inadvertent!

Most services do not have the option of avoiding a JIT or near-JIT system–either they achieve a JIT production system or they lose customers. This is because customers will not wait more than a short amount of time to be served. Service providers must determine what wait is acceptable, and plan capacity accordingly.

A useful way to plan capacity to manage queues is through “queuing theory.” With queuing theory we make simplifying assumptions about service process flows, which allows us to calculate the impact on capacity (service rate) decisions on waiting. The cost of increasing the service rate needs to be traded off against the cost of customers having a long wait in line. Queuing theory helps us estimate those costs.

(Queuing theory was introduced with the Customers in Inventory SBP, and is further discussed in Appendix C. A brief mention of queuing theory is included in the Quantitative Analysis section of this SBP.)

For even more complex queuing situations, we might employ computer simulation to determine the effect of capacity changes on queue performance. With simulation we have a computer model, or representation, of the service process. The model can be “run” to show how the service system will behave given different conditions, such as changes in the number of servers.

What are the major “inventories” in the airline service process? Passengers are inventoried in airport terminals waiting for scheduled capacity (i.e. the airplane). These passenger inventories can be used to handle the delays between various phases of the process: With the hub system, various flights land in a central “hub” airport, where the passengers wait for their connecting flights. The fact that some layovers can be quite long illustrates that this is not an exact JIT system. A exact JIT system (of which few likely exist even in manufacturing) would have arriving flights landing mere minutes before connecting flights. However, the problems of delayed incoming flights would be very costly, hence the need for airlines to inventory passengers for an amount of time in hub airports. But, too much waiting will cause costs for customers.

Another place we see inventories in the airline process is in the queue of luggage going up ramp. Because personnel cannot load and unload luggage until the plane lands, they wind up inventorying the luggage at the gate in a near-JIT manner.

It sometimes seems peculiar that airlines will allow extremely long lines of customers checking in at the gate. Since most passengers have advance reservations, it would seem that airlines should be able to forecast demand for check-ins at the gate quite accurately. Then, queuing theory could be used to estimate how much capacity is needed at specific flight times in order to keep lines small.

With manufacturing, the customers have no idea how long a product or the components of the product were in inventory, nor do they usually care.

1. Where are customer-inputs delayed in the system?
2. How long are they delayed?
3. Why are they delayed?
4. What is the result when delays occur?

The details of this Service Business Principle indicated that queues exist because of a temporary mismatch between capacity and demand. If capacity and demand were better matched then the service could operate with minimal waiting. Describe a place in your service process where customers or their inputs wait to be served. Are there ways that demand and capacity could be kept more in line with one another? List a few ways that capacity could be temporarily increased for those times that demand is high. Also list a few ways the excess demand at high customer-waiting times could be shifted to times when customers would not have to wait. Would a “kanban” system be appropriate, in which the service provider signals to customers when it is a good time to present their inputs for wait-free service?