Tag Archives: quality control

When PDCA Meets Silos

PDCA – Plan, Do, Check, Act (or Adjust) — is one of those acronymic concepts that regularly finds its way into Lean discussions. Descended from Francis Bacon’s scientific method (hypothesis, experiment, confirmation), PDCA has become a ubiquitous catchword for business process improvement.   From standardization and problem solving on the front line to iterative product and process design to Hoshin,  this approach is the engine for continuous improvement.  But like many Lean concepts, when layered over a traditional organizational structure, PDCA can fall far short of its promises.

My initial exposure to the concept, Shigeo Shingo’s Zero Quality Control: Source Inspection and the Poka-Yoke System offered an unusual, non-technical insight into PDCA.   Referring to the concept in the context of quality improvement as “informative inspection,” Shingo posed a  couple of critical questions:

  1. How rapid is the feedback? and
  2. Who is involved?  

Traditional feedback loops were gated, according to Shingo, by a Quality Control function,  a group of subject matter experts “enshrined on a lofty mountain” far away from the “Production Village.”   Several outcomes of this approach were:

  1. Checking (inspection) was a batch process, separate from production, with all of the batch’s attendant delays.  Information was yesterday’s news by the time it reached the lofty mountain.  Whatever conditions may have caused a non-conformance were lost in time.
  2. The person’s doing the Checking were remote from the workers, both physically and interpersonally.   Division of labor became implicitly unequal: thinkers and doers.  
  3. The Doers in Production Village, no longer had responsibility for quality and often no longer had even the capability to Check.   

Regrettably, these outcomes noted by Shingo in 1985 are still commonplace today. As a consultant, I regularly observe long delays to set-ups caused by remote first-piece inspections and worse – forensic root cause analysis initiated long after defects are created.  But worst of all, the folks closest to the  problems are not at the table.  When PDCA meets silos, it too becomes siloed.  Information from production to QC flows through a semi-permeable boundary,  one-way at best and subject to bias and conjecture.  Not a very favorable environment for problem-solving.

Similar boundaries between production and engineering also obscure opportunities for process improvement. In a social model where production workers are doers and engineers are thinkers, the most critical process information is often lost.  An engineering manager once remarked to me “If all employees were engineers, we wouldn’t need mistake-proofing.”  Shingo spoke to this kind of silo as well, coining the term “table engineers” to describe engineers who just sat around a table to solve problems – no interaction with the floor.   These kinds of social barriers dwarf the technical challenges to effectively applying PDCA. 

At the executive level, strategy deployment often only feeds forward only and then typically only to middle managers.  In this case, the silos are vertical as well as horizontal.  Eli Goldratt likened this approach to a game of chess where the players were in a different room from the chessboard and can not see their opponents’ moves.  Check and Adjust steps are not even possible.  And the Doers — employees who must implement  — are frequently not even aware of the big picture. Small wonder that the deployment aspect of strategy deployment is frequently lackluster. 

In fact without acknowledgement of traditional organizational boundaries and application of intentional feedback loops,  PDCA can be short-circuited between any two disciplines yielding only the appearance of science.    The problem to solve is not technical.  As Steve Covey noted,

“A cardinal principle of total quality escapes too many managers: you cannot continuously improve interdependent systems and processes until you progressively perfect interdependent, interpersonal relationships.”

Where are your PDCA boundaries?  Are they barriers or intersections?  How are the interpersonal relationships?   Do pecking orders short-circuit PDCA?  What systems do you employ to foster the free flow of information?    Please share a  thought.


PS I don’t know about you, but I personally am looking forward to “Getting Back to the Future” and seeing old friends and new, at the 17th Annual Northeast Lean Conference – LIVE & IN PERSON – in Springfield MA on October 6-7, 2021. Registration is open and there’s an early bird rate in effect until the end of July. Trust me, it’s a bargain. There will be four tracks, four super keynotes, dozens of presentations that will educate and inspire you and your whole team, plus benchmarking in the Community of Lean Lounge and networking at our Lean After Dark social event. Will you join us?


crackedAccording to the USDA Egg Grading Manual,  “Checks [aka ‘cracks’] are an unavoidable problem in the marketing of eggs because eggs cannot be assembled, graded, packed, transported, and merchandized without some breakage. “   Unavoidable.  That’s the standard, I guess.   The grading manual does not cite a specific AQL for cracks, but clearly it implies some number above zero.   According to ISO standard 2859,  AQL, Acceptance Quality Limit, is the “quality level that is the worst tolerable.”   ‘Tolerable to whom?’ I ask.

Who buys eggs without opening them to check for cracks?  Because the industry has decided that cracks are ‘unavoidable’, we, the customers, routinely inspect for broken eggs.  Most persons I meet, outside of a few in purchasing or quality, have never even heard of AQL or the statistics behind it, but all are routinely subjected to its outcome.  I first became aware of AQL in an oblique fashion when I worked in an IT department attempting to implement MRP.  One of about two dozen order modifiers provided in our MRP software, AQL, was the means by which our buyers enabled suppliers to pass inspection for incoming material with defects.  If incoming material could not pass a 1% AQL, it might be tweaked to 1.5%.   We unfortunately decided that inspecting defective lots was sometimes  more tolerable than running out of parts.  After a time, it became our standard, just like checking for cracked eggs.

For the last decade, more or less, I’ve carried an egg carton with me to customers as a prop.  The message attached to the egg carton  is this:

“The only acceptable level of quality from the customer standpoint is zero defects.”

This fundamental principle behind Shigeo Shingo’s zero quality control became a turning point for me in my own understanding.  As customers we should not consider any other level ‘tolerable.’  Once I seriously adopted this principle as a customer, our suppliers became better suppliers  — and we, too, became a better supplier to our customers.

Restated from a supplier’s perspective we say “Never pass a defect.”  That’s the ideal condition for our customer.   To emphasize this thinking, I’ve returned once more to the carton of eggs with a short video vignette about passing defects.   It’s taken from a new GBMP DVD entitled True North In A Nutshell.  Have a laugh: http://youtu.be/MQrB-HIJWlk .  (When was the last time you had egg on your face?   Share a story.)

We’ll release the True North DVD at our Northeast Shingo Conference in Hyannis, Massachusetts, September 24-25 … less than two weeks away!  Hope to see you there.


BTW:  My next free webinar, “Tea Time with the Toast Dude”, entitled Managing Up (click to sign up) is coming up next Tuesday, September 10 at 3:00 p.m.  I’ve had many requests to weigh in on this subject.  And one lucky participant will win a free registration to our Northeast Region Shingo Conference.