LeadershipLeanSix Sigma

Going to Gemba vs. Statistical Analysis

By Ron Pereira Updated on May 18th, 2017

Early in my professional career I worked as a Process Engineer at Nokia (in their mobile phone division).

I had many responsibilities but, ultimately, my most important job centered on ensuring all of the equipment was producing cell phones that worked the first time.

Go to Gemba

go to gemba
We lean practitioners often preach about the importance of going to gemba, or the place the work is done, in order to understand what’s really happening.

I’m happy to report that I did a pretty good job at this as a young process engineer. You see, I had no choice; my desk was on the production line!

This meant if a production operator had a question they could turn from their machine and be at my desk in seconds.

This also meant if an andon light, or buzzer, went off I was close and could respond immediately if needed.

And because I was constantly at gemba, I knew the processes.  I could hear, and many times feel, when something wasn’t right just like a parent can hear or sense when their child isn’t feeling well.

And, most importantly, I was able to build a strong bond with production operators, supervisors, and technicians.  We were like extended family members all working together.  I still consider many of them friends to this very day… I even got a LinkedIn message from one yesterday as fate would have it.

So, to be sure, “working at gemba” had tremendous upside.

Statistical Analysis

CCD DOEAnother aspect of my job focused on preventive maintenance and built in quality.

As such I was constantly performing Process Capability studies across all of the equipment to ensure things were running smooth.

We also used extremely high tech optical inspection equipment to examine the circuit boards as they moved down the line.

Well, as with any sort of measurement equipment we needed to perform MSAs (Measurement Systems Analysis) to ensure the equipment was working in a repeatable and reproducible manner.

We also leaned heavily, no pun intended, on Statistical Process Control in order to alert us of any special cause variation that may have snuck its way into the factory.

And, finally, I performed hundreds of DOEs during my days at Nokia.

Many were small, 2^2 full factorial designs, that took very little time to complete. But I also performed some pretty awesome, and rather complex, Central Composite Design (CCD) DOEs in order to really dial a particular process in.

Best of Both Worlds

So, the point of this article is simple. I was successful at my job because I worked at gemba and also knew how to do more advanced statistically based analysis.

I could have stood at gemba all day long and watched our processes spin out of control without a clue as to how to help. I could have even asked why a million times and, sometimes, never identified the root cause of the problem.

Likewise, I could have sat in an isolated office somewhere studying my Cp and Cpk results from afar and sounding an alarm when something went wrong.

Had I done either of these in isolation I’m convinced I would have failed miserably at my job.

But, because I did both – lived and worked at gemba and leveraged statistical analysis – I was able to do very well. I also had a tremendous amount of fun.

You Don’t Have to Choose

Sadly, I often sense so-called “lean folks” feel as though they’re selling out if they use or learn statistically based tools.

And I also sense so-called “six sigma folks” feel as though they’re too good or too smart to spend a lot of time at gemba. Instead, they feel like they can perform some Minitab magic and solve all of HRs problems without talking to a single person.

Both of these “purist attitudes” are dangerous, wrong, and extremely short sided in my humble opinion.

Honestly, I could care less what you call yourself. Call yourself a Black Belt if you want… but make sure you’re at gemba more than your office.

Or call yourself a Lean Practitioner or Lean Consultant…. but make sure you know how to perform basic statistically based tests since being at gemba to see defects running off the end of the line may not be much help.

As continuous improvement practitioners the good news is we don’t have to choose. We can learn everything there is to know about making things better.

Do You Agree?

Do you agree with me?

Perhaps you’ve seen folks constantly at gemba without a clue as to how to help or perhaps you’ve seen people whipping up all kinds of fancy charts and graphs but never stepping foot onto the shop floor.

No matter the situation I’d love to hear your thoughts. What do you think?

  1. Abe Nelson

    May 6, 2013 - 7:18 am

    lol, I see this every day of my life on both side. The stats guys sit in cubicles and hammer away on keyboards while the IE’s run around the factory moving rubbish bins into newly optimised locations but have very little impact to the business. The true shame is both groups could probably add value if they would find a way to work together.

    • Ron Pereira

      May 6, 2013 - 8:11 am

      Thanks for the comment, Abe.

      An even better situation may be when the Industrial Engineers and Quality folks walk in the other person’s shoes for a week or two. Sometimes learning what other folks do, on a daily basis, can be especially powerful.

      Then, like you say, the two groups may learn to work together in harmony.

  2. Vishy Chandra

    May 6, 2013 - 8:30 am

    I have been an avid fan of your writes on LSS Academy. Yes you are right. Often, the SS practitioners (Most of them misled) say that GEMBA walks and Genchi Genbutsu are not as worthwhile as their snazzy Minitab graphs.

    A mix of both is indeed right. What I have done most often is let the data and statistical analysis do the talking. When statistical analysis doesn’t point me to a solution/root cause, I take the GEMBA way.


    • Ron Pereira

      May 6, 2013 - 8:53 am

      Thanks for the kind words, Vishy and I’m glad you’re finding a way to do both!

  3. Eric

    May 6, 2013 - 8:33 am

    Ron, well said. I’ve had a similar set of experiences in my career and do not see a need to choose. But among those who do see the need to choose it appears to be either:
    1. standardizing the approach to problem solving prevents people chasing tools rather than solving problems
    2. Ideological fealty to one discipline or the other “Doing LEAN” or “Doing Six Sigma”

    Situation 1 can be a legitimate reason to perhaps suggest keeping things the same for everyone (so long as it doesn’t get in the way of solving problems and improvement). But I’ve encountered more than a couple of situation 2 which can be a depressing place to work for, near or with.

    I think the problem is one of education; thought leaders on the value of tools doesn’t have to upset the approach to the organization and that we’re here to help improve not necessarily push our wares. But also puts an increased burden on practitioners; we can’t dabble in each and master neither lean or six sigma but nor can we turn a blind eye to one discipline. Physicists take Chemistry classes and vice versa.

    Great post, we need more of this line of thinking out there.

    • Ron Pereira

      May 6, 2013 - 8:56 am

      Hi Eric, yes, your #2 is probably the one aspect of continuous improvement that saddens me the most.

      I’ve long tried to understand “why” people become so passionate about why “their” approach to solving problems is better than others… but, alas, this is one problem I have yet to successfully counter with 100% effectiveness!

      Thanks for the comment!

  4. Robert Drescher

    May 6, 2013 - 12:41 pm

    Hi Ron

    I agree with you 100%, we all need to know and understand the work place we have to spend time on the gemba, but we should also be willing to use any type of tool that can helps us and our organization do a better job. SPC can be a great help even to an extremely good company. I know of an example were they have not shipped a bad part in years, but they use SPC to help monitor their process and the equipment as it sends them warnings before anything ever puts them out of spec. It is nuts to try and live in isolation in any one way of doing anything, life is complicated, and we should use any tool that can help, whether Lean, Statistical, or anything else.

    • Ron Pereira

      May 6, 2013 - 12:50 pm

      Thanks for the comment, Robert.

      Your comment made me remember the time I visited a company that produced Integrated Circuits. Their manufacturing process was hyper sensitive… down to operators wearing make-up could create all kinds of production problems. This company also used advanced SPC to monitor every aspect of their business. It was extremely cool to see it in action.

  5. Ed Anderson

    May 6, 2013 - 5:09 pm

    In my world of construction we run into the same thing when superintendents say “My foremen belong in the field, not in the office planning work.” It is difficult to strike that balance between being in the field/gemba and being in a Big Room planning production.

    • Ron Pereira

      May 6, 2013 - 5:19 pm

      Thanks for the comment, Ed. Are you able to practice lean or other continuous improvement methodologies on a regular basis?

  6. Mark Graban

    May 7, 2013 - 8:04 am

    I think a mix of methods works best. In place of statistical analysis, we could simply ask employees why they have left… but exit interviews are notoriously inaccurate for many reasons. You might get bad data… or bad “facts” from what people say on the way out the door.

    I like the Taiichi Ohno-ism of “Data is of course important in manufacturing, but I place the greatest emphasis on facts.”

    I’d also be troubled by the countermeasure of “don’t hire people who live too far away,” as that could inadvertently screen out a great candidate who might then decide to move closer to the workplace or it could open the company up to accusations of discrimination.

    • Ron Pereira

      May 7, 2013 - 9:02 am

      Thanks for the comment, Mark!

  7. Ken Cook

    May 7, 2013 - 8:27 am

    Ron, I will add to the chorus of people that whole heartedly agree with you. As someone who also lived on the production floor, I found that being able to walk in both the hands-on world and the analytical statistical world allowed me to identify and crack problems that eluded others. I find that statistical analysis can provide insight into “where” and “when” to ask questions, and to confirm the impact of a change or an experimental treatment, but rarely do the numbers tell you “why” something is happening, “how” to go about effecting change, or even judge if a statistical artifact is significant to the business or the customer. Only going to (or preferably living in) gemba can provide context and help to answer “why” and “how”, and perhaps more importantly, allow you to leverage the insight of the people that live the process every hour of every day. Ultimately you need both.

    • Ron Pereira

      May 7, 2013 - 9:02 am

      Thanks for the comment, Ken! I am glad to hear you’ve had success doing both!

  8. Joe Brennan

    May 7, 2013 - 8:49 am

    Ron – great post.
    I have been embarrassed so many times when I try to use one work style alone. I have learned to ask rather than tell – since I often get it wrong when acting as the expert.

    The people that run their process have a lot to tell you – if you listen.

  9. Ahmed

    May 7, 2013 - 4:14 pm

    Totally agree Ron – The presupposition that one is superior to the other is flawed. Both are complimentary to one anohter and for my experience truly enhance your understanding of the problem.

  10. Wes Caroland

    May 7, 2013 - 6:16 pm

    Correct root cause analysis almost always makes sense to everyone. It is common sense, obvious as all-get-out, once you’ve seen/heard it, and statistical analysis didn’t make it that way. It more often than not is ‘discovered’ at gemba.
    On the other hand, statistical analysis blows its own trumpet when best corrective action strategies are proposed. This the time to let the data do the talking.
    So gemba and statistical data have their appropriate uses.
    Screwdrivers for driving screws, chisels for chiseling, and files for filing. Never mix them up.

  11. Jessica M. Castillo

    May 7, 2013 - 8:27 pm

    Yes, I agree. I could relate to your experience and your views.

    What is more important is not the name or the approach, but the outcomes.

  12. frans moelands

    May 8, 2013 - 12:29 am

    Every problem is different, somtimes you have to see it on the shopfloor sometimes statistics will get you in the right direction.
    Statistics are great to confirm your feelings but there is always a great risk that you want to see things in the graphs and plots that actually are not there.
    Real insight in the process, Gemba, is needed, to understand the statistics.

  13. vivekanand

    May 8, 2013 - 4:51 am


    this is vivekanand from India,

    what you are saying is 100% right. I am black belt in Lean and Six Sigma, even though the company in which I work believes only LEAN, i learned six sigma out of my know interest.

    now i apply both the concepts for bettering the process

  14. Kicab

    May 8, 2013 - 6:17 am

    Ron, I agree with the gist–why choose? I’ve been doing the “same thing” for more than three decades–process improvement and problem-solving. What people call what I do and my title have changed various times.

    When I hear people say they do lean or six sigma (or lean six sigma), I wonder why they think there are only two choices. Sure, it’s step up from viewing the world as only one or the other, but it’s still myopic. There are many gray shades in between–and other colors!

  15. Alan V

    May 8, 2013 - 11:35 am

    I didn’t even realise this was a practice that I had adopted. Many years ago I realised that being at the centre of the production unit provided a vast field of information. Not only producing a strong product/machine knowledge but it also provided an excellent channel for communication. Liaising daily with the people at the sharp end of the manufacturing process enabled a mutual trust to develop and ultimately many solutions to long standing issues. Discussing quality improvement tools and methodologies (DMAIC, DOE, SPC etc) has also developed many of the workforce, and what they once feared has now become a “friend”. Talking to the person about issues can often be more productive than posting several charts and graphs from a “faceless” person in a non-descript office. I will certainly share the term and philosophy of Gemba within my organisation!

  16. Prasad

    July 1, 2013 - 11:36 pm

    Ron, It is great Communication.

    One clarification. How do you process capability of an equipment like Heat Exchanger ?

    Thanks for your response.

  17. Yves Roy

    July 4, 2013 - 9:16 pm

    I do agree with you, since I am a continuous process improvement and innovation consultant. You have no choice to use the appropriate tool within the right context. From the begining in 2005, I started using Lean and Six Sigma. In public service, Lean approach or the lack of use of Lean approach provides the solutions or the factors that would reduce variations in the operation.

  18. Christopher Thuraisingham

    July 28, 2013 - 8:27 am

    Don’t choose, use the best principles of both worlds for sure! Nice article.

    Thanks for sharing!


  19. Christopher Thuraisingham

    July 28, 2013 - 8:30 am

    Nice article. Yes it makes absolute sense to blend the best of both worlds, isn’t that why we learn ‘Lean Six Sigma’ anyway, since they both compliment one another so well? 🙂

    Thanks for sharing…


Have something to say?

Leave your comment and let's talk!

Start your Lean & Six Sigma training today.