That Day, When the CPK Report Came Out, the Entire Room Fell Silent for Three Seconds
Do you remember? A few years ago, our factory introduced a batch of new machines, touted as highly automated, promising sky-high yield rates. And what was the result? After a month of trial production, the first CPK report came out: 1.08. The meeting room instantly froze; everyone exchanged glances, some even secretly gasped. To be honest, this number was even worse than what we achieved with our old machines operated manually! The boss's face turned green, and he directly asked, "So, we spent a fortune buying a pile of scrap metal, and now we need to hire more people to look after these 'automated' machines?" At that moment, I thought, this is embarrassing; automation isn't a panacea.
Where Did the Problem Lie?
To put it bluntly, many times when we pursue automation, we haven't clearly thought through "what should be automated" and "what shouldn't be automated." As soon as people hear about automation, they perceive it as high-end, a future trend, and then blindly assume machines should replace humans in all steps. But the reality is that some processes are simply not worth automating, or rather, the benefits brought by automation cannot offset the invested costs and maintenance hassles. Do you really need to spend millions to buy a robot to move something that's only moved three times a day, two kilograms each time? This is the "Manpower vs. Automation" decision matrix; the key consideration has never been "can it be automated," but rather "is it worth automating."
How to Do It in Practice?
Frankly, to judge this, you can think from several dimensions, like a four-quadrant matrix:
- High or Low Operation Frequency: Actions repeated hundreds or thousands of times a day should, of course, be prioritized for automation. If an action is performed only once a week, the automation benefits will be very low.
- Degree of Variability: If the results of each operation are almost the same and predictable, robots are happiest doing it. However, if adjustments are required for different situations each time, human judgment is far superior to machines.
- High or Low Cost of Error: If a small mistake could lead to losses of hundreds of thousands or millions, then automation to reduce the probability of human error is absolutely worthwhile. For example, in wafer handling, dropping one means it's a complete loss.
- Manpower Costs and Skill Requirements: This is the most direct. For highly repetitive but attention-intensive tasks, such as visual inspection, employees might suffer occupational injuries or turnover rates might be high; in such cases, automation can solve many problems.
For example, in our factory, one process is "screw fastening." We need to fasten twenty thousand screws a day, and manual fastening often results in a DPMO of 6210, as issues arise if screws are too tight or too loose. This task has low variability, high frequency, high error cost (ruining one board means it's scrapped), and workers easily get fatigued. In such a situation, introducing an automatic screw fastening machine, even if it costs millions, is cost-effective in the long run. But if you talk about automating the process of "meeting with clients," then I think you might have misunderstood something.
The Most Common Pitfalls
The pitfall we most frequently stumble into is seeing only "automation saves manpower" while overlooking maintenance costs, failure rates, and flexibility issues after implementation. Once, to save three laborers, we purchased a highly complex automatic assembly machine. The result was that the machine broke down frequently; each time it stopped, the entire production line would halt. Calling the original manufacturer for repairs was prohibitively expensive, and the back-and-forth would take two to three days. In retrospect, the money saved from those three laborers wasn't even enough to cover the repair costs and downtime losses. Honestly, if we had used the money saved to buy better coffee machines and improve the working environment, employee efficiency might have been higher, and they wouldn't have had to bear so much stress. So the key point is that automation is not as simple as just saving manpower; it also includes the hidden cost of "reliability."
One Thing You Can Do Today
List out the processes you are responsible for, choose one step, and evaluate its "variability" and "frequency."
Article Category: Lean Production