Lessons from Japan: Rapid, Disciplined Problem Escalation

Note: Links to all of four Japan factory tour posts and the various lessons from those tours can be found here.

One of the most memorable parts of my trip to Japan was watching the Toyota Kyushu manufacturing lines, and especially how the auto assemblers handled problems.

If the operator has a problem, he pulls a cord that sounds a chime.  Another person in white gloves comes running.  The problem escalation method is something many of us struggle with, and the Toyota solution is to create very simple decisions and multiple decision points.  The operator begins with "is there a problem, yes/no".  If "yes" he pulls the cord.  The guy in white gloves comes running and has another simple decision "is it really a problem, yes/no".  The two of them have literally under a minute, one takt, to make that call.  If "yes" then the line stops and the supervisor comes running.  He has a simple decision of "can the problem be fixed within two takts (about two minutes), yes/no".  If "no" then the decision goes one more level to "shut the factory down".  Within three minutes a problem has been identified, attacked, almost always resolved, but potentially a decision has been made to shut down an entire factory.

How often does the chime sound?  About once a minute throughout the factory.  There is no problem reporting a problem.  The intent is to find it and get it fixed, fast.  As in within a minute.  How long does it take you to identify, report, and fix a problem?  I know what you're thinking, I'm embarrassed too.

There are about seven parallel assembly lines, running head-to-toe, to make each vehicle.  Each are synchronized but independent to allow for a three to seven vehicle buffer, which thereby allows for each individual line to be shut down for two takts two or three times per shift without shutting down the entire factory.  "Buffer" is probably a misnomer... remember that it represents about five minutes.

It really was happening literally every minute.  The chime, the running around, occasionally the line would stop for a takt, then it moves on.  Problem identified, investigated, and resolved in under a couple minutes.  Longer than that and a decision is made to shut down a $1 million per hour factory.

Pressure?  Not really.  The decision-making is so defined that it becomes easy.  The desire to find problems easily outweighs the potential cost.  Shutting down the entire factory is still better than letting a customer receive a quality problem.

How many of us get squeamish over a $5,000 problem, or a decision to shut down a line making a few grand an hour?  How many of us try to rationalize issues, subjectively decide if they are "acceptable" or "noticeable" or try to "reinterpret specification?"  How much time is wasted when a line is "almost ready" to start producing after a changeover?  Why do we let that happen?  We're all focused on quality and customer value, right?

Because we haven't made the decisions easy, and high impact decisions acceptable.  Because we are still concerned about cost over quality, not realizing (or accounting for) the fact that the eventual cost of poor quality far exceeds the cost of finding it early.

The key aspects of the Toyota system are:

1. Very clear and discrete decisions at each level: typically "yes/no"
2. Very defined escalation structure from line operator to support guy to line supervisor and above
3. Very defined time intervals in which decisions have to be made
4. Acceptance and outright desire for problems to be found and action taken, sooner rather than later
5. The root cause of problems is investigated and rectified so they don't occur again

That doesn't sound too hard, does it?  Top leadership must reinforce and support the identification of problems without repercussions. The escalation criteria must be agreed to, and people held accountable to executing the escalation structure within the criteria and defined time intervals.  Top leadership must become part of the escalation structure.

Of course not every problem can be solved in under three minutes.  But I bet we'd be surprised how many could be.