Automation can be key to business efficiency, but only if done right. Smart orgaizations know if they first simplify a process before they automate it, this will ultimately lead to reaping the greatest amount of rewards from the automation process.
Certainly, automation can reduce cost, create more throughput and, in many cases, improve quality. Theoretically, any process can be automated. Some are more difficult than others. One key to improving the success of automation is to simplify your process before attempting to automate it. Benefits of simplifying include:
Ease of automation;
Reduced cost of automation;
Easier transition from current process;
Higher return on investment;
It is just plain easier to automate a simple process than a complex process, and the greater the complexity, the greater the difficulty and cost. Complex processes generally require more complex automation, which is more difficult to design, build, debug and install than their simpler counterparts. This difficulty consumes capital investment, lowering your payback.
Finally, once a process is automated, it is usually difficult to change. This provides the benefit of process standardization, but makes process improvement difficult. Generally, improvements are best made before automating. Failure to do so may cut you off from ever getting those improvements implemented.
Types of Processes
Not all processes are equally suitable for automation. We can group most processes into three broad categories: physical, transactional and hidden mental processes. There may be other ways to categories processes, but this will suffice for our discussion.
Physical processes are those that accomplish physical work, such as machining, assembly, transporting and shipping. Transactional processes accomplish non-physical work, such as order processing, scheduling, inventory management (excluding the physical part) and payroll processing. Hidden mental processes accomplish non-physical work by thinking, such as assessment, diagnosis, design, quoting and other processes that are generally thought to require judgment.
Hidden mental processes are usually the most difficult to automate, simply because they are not well understood. They are what goes on in the head of an expert — a doctor diagnosing a medical condition, a mechanic analyzing a malfunctioning machine, a TSA agent reviewing images of luggage to find security threats. These are all processes, but not well understood, even by the experts themselves. These may well be processes worth automating, but until their steps can be identified, they simply are not suited for automation.
Physical and transactional processes are often excellent candidates for automation, though the approaches are quite different. Physical processes are automated through machinery, which must be engineered, fabricated and installed. Transactional processes are typically automated through computers and software. Even so, both types of processes share the common characteristic of being composed of a sequence of work steps, each of which accomplishes a function. The clear identification of those steps must be done before the process can be automated.
Such a well defined process may be then automated, but that can result in excessive complexity and cost, often providing lower performance than needed. No process can be automated without first defining its steps, but the second step should always be simplification.
Simplifying the Process
At the most basic level, simplification is merely a review of the process, removing steps that are not necessary. This can appear almost silly, since why should there be unnecessary steps in any process? Every process, even the best, contains both value-added and non-value-added steps. The value-added steps are those that directly contribute to accomplishing the purpose of the overall process. Non-value-added steps do not. Theoretically, there shouldn't be any, but it always seems to happen.
So, how much non-value-added is okay? Strictly speaking, none of it is okay, but high-efficiency companies typically will have 70% or more non-value-added activity, as measured by time spent in the process. If we think of an assembly process, the value-added steps would be those which actually join parts together. Nearly all other steps, such as transferring parts into position, pre-positioning parts, starting fasteners and transferring the assembly out, are non-value-added.
Many of the non-value-added steps may be necessary, but only because we can conceive of no way to accomplish overall process objectives without them. If we could, it would be a simpler process. In a physical process, there are things that go wrong and often a need to detect and act upon such conditions. Both the detection and the correction are non-value-added because they would be unnecessary if the error could be avoided from the start.
If a product design calls for 18 screws to secure an access cover, of course we need to move 18 screws into position and drive them home, perhaps repositioning the assembly between driving screws. But if the product could be re-designed so that the access cover slid into a channel in the rest of the product, requiring only two screws, that would be a much simpler design to assemble. Simplifying the product that is assembled by a physical process will not only save cost but also improve the profitability of the product itself.
It is not the intent of this article to show how this simplification is done, only that it should be done. The method for simplifying a product was developed in 1947 by Larry Miles of General Electric, known as Value Analysis and Value Engineering (VAVE) and added to by Boothroyd and Dewhurst in the 1960s in their Design for Manufacturing and Assembly (DFMA). These techniques are basic engineering approaches that are all too often overlooked at great cost.
Another technique for process simplification is to eliminate opportunities for errors by mistake-proofing (poka-yoke). Suppose a part that is inserted upside down will assemble but cause the finished product to be defective. We might build sensors into the process to detect such a defect, alarming or causing the defective product to be ejected from the process. All of the steps that would be taken to correct the problem would be non-value-added. It may be more cost effective to repair a minor error than to scrap the assembly, but none of those steps would be needed if the error had not happened. Mistake-proofing prevents the error from ever occurring, so the correction, and even the detection, can be eliminated from the process itself. Shigeo Shingo has developed a host of methodologies for this and has been published extensively.
Transactional processes nearly always contain much waste, but unlike physical processes, the waste is hidden. Defects in a physical process are often very obvious, whereas defects in a transactional process may simply cause work to be corrected somewhere downstream. Transactional processes, by their very nature, are invisible. If you watch workers in a physical process, you will see parts added and adjustments made, and will often be able to see the product emerging. Workers in a transactional process may be equally busy, but all you will see is their dealing with paperwork, typing at a computer or talking on the telephone. The process by which they do their work is not evident and neither is most of the waste.
To simplify a transactional process, we first have to make the process visible. This is typically done by process mapping, also known as flowcharting. By collaborating with those who actually do the work, we document the sequence of steps by which the work is done. Usually, the outcome of process mapping shows the process to be more complex than anyone had imagined. The fact is, in transactional processes, simplicity happens as the result of simplification efforts, while complexity happens all by itself. Over time, transactional processes change and steps get added, but there is usually no explicit effort to eliminate steps that are no longer needed. As a result, work that is no longer needed may continue for decades.
For instance, a shipping process in a Michigan furniture manufacturer had more than 200 steps. Leadership was amazed that something as simple as putting product in a truck could be so complex. Of all of these steps, very few involved touching product, but rather processing information (packing lists, bills of lading, customs documents, etc.).
A transactional process may be simplified by carefully reviewing each step and determining the value-added steps. In most transactional processes, there are typically less than 10% value-added steps. All other steps are candidates for elimination. One of the major opportunities can be found in error-correction loops. Just as in physical processes, things go wrong in transactional processes. When they do, there is an inspection that finds them and a set of corrective steps. These loops cause processes to become larger and larger over time.
In the shipping process above there might be a point at which it is discovered that a valid “ship to” address is not known, prompting a flurry of activity to determine the address. Rigorous examination of the process might reveal that the missing address was caused by an oversight in the order-entry process, which takes place in an entirely different part of the organization. The appropriate action would fix the order-entry process in a manner that guarantees that the “ship to” address is always correct and complete. With this done, the shipping process would never have to deal with the error, and the process would be correspondingly simpler. Most transactional processes use computers and software for at least a portion of the process. Software provides a good way to mistake-proof the process.
Anyone who has bought things online has experienced this. Most well designed merchant Web sites will take you through a series of steps and simply not permit you to proceed if there is missing or invalid information. Ensuring that e-mail addresses fit the usual format, credit card numbers are valid and “ship to” addresses are entered (if different from the billing address) are examples.
Nearly everyone has been or will be touched by automation of some sort. If you are party to a process that is being automated, be sure to ask the question: “What simplifications have we made in the existing process?” Hopefully, it will be the beginning of a discussion that will result in substantial improvement and lower cost. Sometimes, you'll find that after simplifying a process, there just isn't enough left to make it worth automating. That's a bad deal for those you would hire to automate, but a very good deal for you and your company.
William (Bill) Eureka is president of consulting firm EurekaResults.com, headquartered in Lowell, Mich. He has more than 40 years of experience in engineering, manufacturing and consulting with more than 400 companies. He is a Six Sigma Master Blackbelt, Jonah (Goldratt Institute, Theory of Constraints) and Lean Master, with more than 60 kaizen events completed.