As everybody knows by now, in China labor costs have been rising quite fast over the last few years, and this trend is here to stay.
What is the typical Chinese response to higher labor costs? Hire less employees, of course. And how to hire less workers? By automating factory operations.
Unfortunately, by pursuing full automation, they do not address the most important opportunity for savings – the optimization of the whole system. I already wrote on this subject two years ago. David Levy agrees with me.
I am not suggesting to do everything manually, of course. The right answer is often semi-automation.
Let’s take a simple example. According to Wikipedia, a semi-automatic firearm performs all the steps necessary to prepare the weapon to fire again after firing.
What is the different with automatic weapons? Semi-automatic firearms do not automatically fire an additional round until the trigger is released and re-pressed.
In other words, it makes a certain task much easier and faster, but a human still needs to take an action.
In office work, semi-automating a task might mean designing a point-and-click operation on a computer. The program would then do the heavy work, while the human operators spent only half a second on this task.
In a factory, a small machine and/or special tooling can often do the job.
Benefits of semi-automation:
1. Labor productivity is greatly improved
A process that is not automated at all can be very time-consuming. The jump from no automation to semi-automation brings huge benefits, and the jump to full automation often brings limited benefits.
2. Human operators still need to use their brains
They still “pull the trigger”. It means they have an understanding of the steps to follow, and they can think of better ways of organizing their work.
It also means they can work on more tasks and have a richer work content (I bet Foxconn workers wouldn’t jump from rooftops if they could learn something on their jobs).
3. Processes are much more flexible
Since each task (rather than the whole process) is automated, the general workflow of a process can be reorganized easily.
4. Continuous improvement is possible
Once you buy a big machine, you are stuck with it. Continuous improvement is impossible. The key is to keep control over the sequence of the tasks and the way they are performed. It is possible with low-complexity semi-automation.
So, when is full automation preferable to semi-automation?
When human intervention is not a good thing
I was told that engineers at Boeing and Airbus dream of removing pilots from their planes. Their reasoning is that most crashes are the result of human errors (because of panic, stress…). If that’s true, maybe an IT system is preferable to a pilot.
Another example: if I am driving a car and it loses its grip on the road, I am happy if the car’s IT system auto-corrects the trajectory. The human operator is not fast enough to do the necessary adjustments.
When we don’t think a process can/should be optimized
There are cases where a process is unimportant and should be fully automated. For example, a thermostat system can start heating when the temperature drops below a certain limit, without input from a human.
However, this is not applicable to business processes that are at the core of a company’s activity. Go for semi-automation whenever possible.
What do you think?
Renaud, spot on. This is one of the things we tell our clients. The most effective companies are those that implement some level of automation but keep also a level of flexibility. This is one of the strengths of Chinese suppliers for mid size buyers: flexibility that makes high mix – medium size order still interesting for the supplier.
Of course, every time manual work and machine setup is involved, the routine QC process in place is critical in order to ensure consistent production.
The the right way to go is semi automation with solid QC and processes.
You are right, it also allows for a wider product mix. Good point.
Renaud, as usual you nailed it. US companies sometimes go overboard on automation, introducing robotics to perform tasks that humans perform more efficiently. The key is to break down the tasks required and let humans do what they do best (acting with insight and flexibility) and let robots add value where they can (usually when high accuracy is required in a highly repetitive fashion, or where conditions are dangerous, uncomfortable or unhealthy for us primates).
The next step after semi-automation is, in my thinking, to widen the range of tasks each operator performs, putting him/her in control of as many process steps (automated or not) as possible. This “cellular” approach achieves seamless process flow, along with much better operator satisfaction.
I’ve often felt, as you seem to, that a large part of Foxconn’s problems stems from the repetitive nature of the work their operators are required to perform.
I can’t agree more! Thanks David for adding to the article.