Training Operators to Follow WIs

I am currently working at a company that has yet to implement any Work Instructions. I've been creating them based on operator input to help determine the best process that we're currently aware of (watch operators of each shift do work, take pics and notes, create rough draft, have operators review, etc).

I'm supposed to train the operators on these Work Instructions that they have helped me develop, but I'm having trouble trying to figure out how to teach the importance of following the same sequence every time. I explain what we're trying to accomplish, to have everyone performing the work the same way to help create a consistent, quality product while making it easier to figure out how or why something went wrong if a problem is spotted. I also explain how during shift changeovers, if everyone is following the same sequence, an operator can pick up from where the previous operator left off without much trouble.

But when it comes to operations that can be done at multiple points throughout the beginning of an operation so long as it's done before a particular step (ex: ensuring that a sharp razor is loaded into the knife before making a cut), I find it hard to argue the question, "Well, as long as I make sure it's sharp before the cut at Step #7, what does it matter if I do it Step #3, Step#4, or Step #5?".

Do you guys have any advice, or notice that I'm not explaining something as I should?


Variation in the way a process is performed can impact both upstream and downstream processes, for a number of reasons. This variation can contribute to your not making Takt time, which can affect your flow time, which can then lengthen your Lead time, resulting in the possibility of the customer not getting their products when needed.

Also, if you agree with the example you wrote regarding the knife, what is the next step that they will challenge you on, and ultimately where will it end? My bet is that it won't. Varying from the current standard is ok as long as it is agreed upon by those doing the work, is documented, and it becomes the new Standard Work.


As any process on production require continuos improvement, the same apply for your WI.
Take note or make a database for any change/suggestion for improve this sheets.

Here on my job, normally the WI/OPS (operations instructions) are audited for AS9100.
So need be as clear as possible and identified with P/N, blue print REV, internal REV. etc.

Picking one way, one standard, leads to less variation in process output. When there is more than one, just collectively pick the best and follow it.

But if the item truly does not need to be standardized, leave it be. Avoid overprocessing procedures.

I really liked your input.

Would you (or other readers) care to elaborate on what processes, tasks, etc. typically need to be standardized (and why)?

And, which ones do not need to be standardized (and why)?

I really liked your input.

Would you (or other readers) care to elaborate on what processes, tasks, etc. typically need to be standardized (and why)?

And, which ones do not need to be standardized (and why)?

It sounds like you've done the right approach - let the operators define the process, explained the major benefits. Does your standard workflow show the benefit of doing steps in a specific sequence - it may be as simple as a couple seconds or an ergonomic benefit. Seconds can add up, even if it takes a month or a year.

Three key areas I also stress: training, improving the process, and increasing output.

From a training perspective - if everyone is doing it the same way, including the steps that don't seem to matter exactly where they are done, it is much easier to train new people AND verify they are performing the work correctly. Also, if every employee has the option to do certain things however they choose, there is a possibility of someone developing a "shortcut" that impacts quality, even though their intent was only to go faster and be more efficient.

Process Improvement: With everyone doing the same standard process, it is very easy to see the impact of process changes, and very easy to implement. Standard work is just laying the foundation for the real improvement.

Increasing output: If the work is standardized, it becomes much easier to rebalance work and potentially add one person that only needs to learn a portion of the work to get a nice productivity increase. If everyone is working in different orders, it's harder to rebalance because you basically need to retrain everyone. The natural rhythm that each person was previously working to was different due to different processing steps, and the new sequences may lead to a quality drop simply because it will seem "new" and "different" in the new process.

If all else fails - Ask them to "Just Try". If it doesn't matter, then why does it matter to them either? The answer may reveal a better process. Usually this wins over even the most reluctant.
Worst case - and this is rare - make the decision, tell them to do it, and evaluate if that person is right for the team. I'd rather have a hard worker with average talent than a person with exceptional talent that brings down the attitude and morale by impeding improvement.

To once again quote Taiichi Ohno, "Where there is no standard, there can be no kaizen".

That to me answers your question.


Each process output has key characteristics. In general, effectiveness characteristics are related to output quality. Efficiency characteristics are related to resource consumption.

For each process, what are the key inputs to consistently meet needs at acceptable cost?

Those key inputs need to be controlled. The procedure (along with training, material, workplace organization, etc.) should be defined to apply suitable control of key process inputs.

Make an increment improvement with a procedure. Set objectives for output quality and cost (time). Measure, adjust as needed.


I know a "techno-geek" who demonstrated the benefits of Standard Work by video taping several cycles of a process being performed without standard work and also the same process being formed with Standard Work. He then used a program to show each method simultaneously side by side. It was obvious to the observer which process method was superior.

Focus on LEARNING.

Start with building facts (performance metrics). Share the 'facts' with the team and build your case for following the WIs with precision. Run controlled tests; (1) w/o following WIs and (2) following WIs; with stopwatch, recording wastes, etc. Compliment the controlled tests with Utube videos and training modules (standardized work, 1 pc flow, etc).

Ultimately, the facts should be recorded / reviewed by the team on a daily basis - this will drive ownership of the problem / exploration of a solution. If the team does not buy-in, you need to start the process of 'observing and discussing' or 'catchball'.

When I design work cells, I build the work sequence into the cell design; i.e. part location following work sequence, tools and work areas are optimized by matching the WI sequence. Consequently, it makes it 'difficult' (harder) to work with variation.

Last resort involves mandating the process but this should be avoided.



Hi Evan

Generally good work instructions follow a logical pattern that results in a good job being done. Now you have started the right way by going to the operators and getting their input throughout the process.

But like all good work instructions they are a living and evolving document, thus osmetime we make errors as to what is the most logical pattern to follow. Take your example with two options:

1. If all the steps preceding step seven the actual cut are manual, than the most logical point to check the sharpness of the razor is immediately before making the cut. What I would do is shift it to that point and if needed enforce it being done at that point, after all they helped create the instructions so unless they have a suggestion to improve them they need to follow them, if there is an improvement suggestion than run a test to see whether it is better for the majority. The biggest reason for creating standard work instructions is so any worker can take over effortlessly without having to figure out what was or wasn't done.

2. Some of the steps preceding step seven are done by a machine, so the operator has to wait for it to complete, thus it would be logical that they check the razor for sharpness and perform blade changes while waiting. If that is the case anyone not following the instructions is actually just trying to waste time elsewhere, so unless there are improvement suggestions the rules need to be enforced so any other worker can pick up where they left off effortlessly.

The whole point in creating standard work is to allow any other worker to step into the job and effortlessly be able to take over and know what to do. When people fail to follow them they defeat the whole purpose of creating them. It also shows that they have very little respect for their fellow workers, managers, the company and the customers because they are willing to cause them problems for their personal attitude. No one individual makes a team; rather it takes the co-operation of every member and their willingness to follow the direction of their coaches (managers) if they hope to succeed.

You need to talk with the objectors and either get them inline or replace them, just like on a sports team a great athlete that isn't a team player hurts his team more than they help it, if that wasn't the case any owner could buy themselves a team and win every year.

Good Luck, and Live Lean and Prosper
Robert Drescher

One of your challnges is that you are trying to change culture. Do the Supervisors and the person they report to believe that jobs need to be standardized to the level you want?

I've worked with three companies with 60, 40 and 20 years existence with top managers having over 25 years and supervisors over 10 years experience and they had "limited" exposure and appreciation to "standardization". They might say it is a good idea, but developing and enforcing can be a different story.

If it helps, is it possible to combine some of the steps into broader "elements". For example, in a lab we needed to cut, prep and mount parts to prepare for testing. Different people had different methods of cutting. One day three people were absent and we needed to figure out how to cut the parts, because there was no documentation. This created the need to develop documented instructions.

When they returned we decided to document and discovered that two of the qc techs were able to perform the task making only two cuts. The third person was the retired metallurgist from one of our largest customers (74 years old - part time) and it took him 3 cuts.

The load in the lab did not require a strict adhereance to a stable takt time and we weren't going to get the retiree to change his method and couldn't afford to lose him due to the credibility he brought to our operations

We ended up documenting the one method and called it the "Two Cut" method and then documented the "Three Cut" method honoring the retiree by putting his name in the title . ex. "Ken's 3 Cut". This helped the lab buy into documenting more of their cutting methods. Due to the extreme variety of parts coming in, and the one time nature of many, it was impractical to document all the parts.


A. What is the risk of not having it documented? Was there ever a significant issue created by someone taking over the operation and making a significant number of bad parts at your place? Or is it for hypothetical Preventive Action. If that's the case, sometimes it's better to prioritize documentation activities by waiting until issues happen, then you can apply PDCA and get them to buy into documenting more. (i.e. "wouldn't it have helped if we had these documented / standardized before?" you say with a slight snicker. - if they say "no", you better understand your challenge)

B. Can you combine "steps" into a standardized "element" (i.e. "cut and prep" on your top level "standard" vs every individual step of the cut and the prep"?)

C. Can you document multiple techniques as long as they don't create issues with Quality, Takt or required Tools, Supplies or Inventory?

Changing culture takes time, and you can't change old dogs overnight. I would probably plan on a 2 - 4 year time frame before you really have the standardization throughout the system to your "liking". In the meantime it's 5 steps forward, 2 steps backwards. Stay focused and keep smiling on the forwards, be patient on the backwards and use them as future PDCA learning points.

Good luck - been there.


Thank you guys for all of the replies. I really appreciate them.

On a similar note to my initial question, how do you guys handle auditing on the work instructions? For example:

1.) Who typically performs the audit: the operators' supervisor, quality control, or just an unbiased person (ex: industrial engineer)?

2.) If someone does not pass the audit, what should happen? I would think that the operator should be immediately re-trained as per the standard, then re-audited afterwards to ensure that the operator has it down. I do not believe that an operator who failed an audit should be able to continue working on production on his own without first being re-trained. If there's no one available for him to train with, he doesn't work production on that operation until the re-training is complete. Am I being too harsh, or am I effectively starting to change the culture regarding how important following the standard is to the company?

3.) How does one "fail" an audit, anyways? I assume how strict an audit will be depends on the company and their own tolerances, but I may be wrong. Also, how strict can it get? For example, if an operator does Step #4 first, then Step #3 (either of which are independent of each other but need to be done before Step #5), then that person still didn't follow the standard. Would that mean that the operator failed the audit?


If there truly is multiple scenario steps to the process, I would run through each one and make sure everybody's idea is worth considering. Analyze each different process and see if it truly makes a difference.

Sometimes you will find that there really is no true standard needed and all your doing is over optimizing the process at hand.

The goal should be that when you have reached the maturity to have an operating rythm, the Standard Work instructions should be maintained and monitored by the owners, the ones that do the work. When you reach this point, periodic checks become unnecessary, as this becomes part of doing the job every day.

In the mean time, periodic checks should be conducted on a regular basis by the process owners and their supervisor to ensure compliance. Notice I didn't use the word "audit". An audit implies that in the event of failure someone is going to get called out. You want to be careful that ensuring compliance doesn't become a hammer that is used to bludgeon people. If you do a good job of teaching and demonstrating the advantages of following standards, the operators should WANT to follow them.


I am not a big fan of "Document-Based" auditing. You might want to read "How to Audit the Process-Based QMS" by Arter, Cianfrani and West to gain a little different perspective on systems design and auditing.

Process-based methodology believes it's more important to manage and control the system than to have a document for every activity in place. Problems with document based methodology include:

A. All documents are given the same weight, when in reality some activities are more important than others.
B. Old practices are institutionalized which can limit improvement initiatives.
C. Individual documents often ignore process interactions and how one activity might negatively impact other activities downstream.

Just because an operator is operating outside of a parameter on a work instruction, why does he / she have to be retrained? Maybe the reason is because there is an issue with the machine or material. Or forgive me, but maybe what's in the document doesn't produce the best results with efficient use of resources.

Auditing and Analyzing a process for improvement involves looking at various aspects of the process including: What are the inputs and who are the suppliers, What controls are in place? How was it validated, and was it proper? How are the processes monitored and measured? How does the process interact with other process? What feedback is there from customers and other processes? How has the process improved? (the book lists others)

In my (and others) opinion, auditing should not be a witch hunt to bust the operators. During the audit it should be up to the auditees to demonstrate what their system is and is it effective. Maybe what is contained in the document is what is bad and the document is what needs to be changed. The re-training might be setting up yourself to get bad parts. Wasn't it Deming who once said, "80%+ of the issues are created by the system, the operator is responsible for less than 20%."? Why was the document created? Why was the operator doing something different? Maybe the system that created the document is what needs to be changes - i.e. maintenance, engineering, purchasing, etc.

Thought this might be of interest, you and others don't have to agree and are welcome to share feedback. Good luck.


Can you give me an example "where there really is no true standard needed and all you're doing is over optimizing the process at hand."

Ken Hunt,

This (above) is what I was exploring in my previous post.


Many times the operator do the operations improving the process and this isn't bad, on this cases the WI need be updated.

This situation happen a lot on my facility, normally the inspectors/supervisor catch this good changes and take note for ECR (ENG change request).

Instead of starting a separate post, I'll just continue on this one. One of the keys as Ken put it is "maturity" of the system. As with the case with many small operations, this example comes from an area still in the early developmental stages of standardization.

Last week we were performing a time study / standardization exercise involving the Wheel Dressing (truing) in a simple Grinding operation. Included in the exercise were two students (approx age 35) who were taking a Lean class at the local community college. It was good having them, as if anything, they pressed issues that could possibly lead to "over-standardizing" the activity as the methodology wasn't as stable as they thought it should have been.

Information related to the activity and the questions:

A. Wheel dressing involves mechanically running a truing device (nib) back and forth across the wheel, to reflatten the surface after the corners have broken down through wear.

B. The dressing process also involves manually moving the wheel in towards the fixture as it has decreased in diameter due to wear. There is no reference (i.e. "micrometer dial") to measure / indicate the exact position of the wheel.

C. Number of dresses can also be influenced by the average depth of the grind. Deeper = more dresses.

D. Not turning in the drive enough, cause the operator to have to go back and turn it in more. Turning it in too far can cause the wheel to crash into the fixture and damnage the fixture. Operators turn in the drive 1 - 4 times per dressing.

E. There are a couple other minor activities involved that are insignificant, such as a method to open the grit on the wheel, but I'll leave those out.

F. The number of passes of the nib is therefore a function of:

1. Number of units made since the last dressing
2. Average depth of the grinds
3. Whether they might "over dress the wheel" which would shorten wheel life.

G. Dressing with less wheel wear, creates more dressings per shift.

H. Wheels are rather inexpensive.

I. Dressing process on an average is about 3 - 4 min with the operators dressing an average of 10 - 16 times per shift.

J. The machines are not linked to the next operation due to environemental considerations. Operators are able to maintain about +/- 5-10% parts produced per shift.

Here are some questions we kicked around related to the extent of standardization

1. On the average the nib made between 25 - 40 passes. To what extent should we attempt to standardize (control) this? +/- 5? +/- 10? Greater? or What difference does it make?

2. To what extent should we attempt to standardize (control) the number of times the operators dress the wheel per shift / number of blades per dress? +/- 10 blades, +/- 25 blades, +/- 50 blades? What difference does it make?

3. To what extent do we attempt to standardize (control) the time to Dress the Wheel. +/- 15 sec? +/- 30 sec? +/- 1 min? Greater? What does it matter?

I wouldn't mind sharing responses with the supervisor, operators and guest students. I'll share my responses to the above if there's any replies that would open the door to such a response.



First of all, great topic & discussion.

I'm of the camp that would say stick to standardizing what is truly critical to the process based on Safety, Quality, and Productivity. That said having alternate sequences of major work elements is a slippery slope. In my experience it is best to fight upfront for consensus on the floor with a flipchart. Then 5S/3P the location of tools, supplies, equipment, work surfaces, etc. around that agreed upon process. Likely picking the best arrangement of these components will be easy with consensus on the sequence & difficult without. In other words shoot to setup the tools, supplies, equipment, work surfaces, etc. so that you would not want or possibly be able to do the process any way, but the agreed upon sequence.

Another trick to getting consensus is to look for an improvement opportunity that galvanizes the group. A machine modification, new tool, Creform build; something that makes their life better AND changes the process for everyone. Then it takes some of the pressure off of just arguing about who has been doing it the best the last however many years.

I have not experienced much in the way of auditing yet. I think there is value in doing it. But I would say there is more value in auditing the process, rather then the person. I have had good results in having operators train supervisors and managers (related to the process at some level) using the TWI's Job Instruction methodology. This flushes out issues with process, the documented standard, and maybe the training methods of the operator.

I appreciate all of the replies, and I've been re-reading them to make sure I understand what everyone is saying...and, to be honest, I'm really confused.

There are people saying that standardization is important, it's a living document, etc., but then also other people saying to watch out for over-standardization. Alright, so...where's that line? How can I tell when I'm going for too much or too little standardization?

And then, as far as the audits go, Ken's saying that the operators should want to follow the standards if I teach and demonstrate to the operators the benefits. To help get insight on this, I brought in an operator who I have a strong relationship with to read what was being said, especially that part. His first response? Laughter. He was laughing at the notion that the operators would actually want to follow the standard. And this is someone who helped me create some SOPs already. I'm clearly not teaching and demonstrating very well. Afterwards, I went over a Lego exercise and the Numbers game with him, and I got him to understand more of what I was talking about, but he's still skeptical of the implementation overall.

So, if I'm not "auditing" but rather doing a "periodic check", how exactly is it done? Watch the whole operation, or just a portion? What happens if a step is missed, or a sequence messed up? If the supervisor is responsible for the check, and the supervisor is "really busy" all the time, how can I be sure that he'll actually check?

I must admit that I am frustrated as I try to decipher between these replies, trying to figure out what's correct and what's not. Just look at the variation in some of the responses. Sometimes...I'd prefer to be presented with a case study or an actual example that I could replicate and learn from on my own instead of reading through varying walls of text.

I have been reading the responses and I must say everyone is right. We all agree in principle standardizing a process is beneficial. We get better throughput, better quality etc. However this reasoning for operators is not enough. Before we start standardizing we need to ask the question what and why do want to standardize. What benefits do we expect or what problem are we trying to solve. Involving operators is critical and you seem to have done this but how do we make them want to standardize??
I've found that the operators themselves need to see the need to standardize.
It is my belief that if the operators have the right KPI's in place, for example our beloved OEE they will seek ways to improve it. Link it to a team based bonus scheme. Once they realize that a lack of standardization is hurting their results they will standardize. This way you will standardize the relevant processes and create a need for constantly reviewing work instructions thus continuously improving the process.

"There are people saying that standardization is important, it's a living document, etc., but then also other people saying to watch out for over-standardization. Alright, so...where's that line? How can I tell when I'm going for too much or too little standardization?"

Evan - Great quote. If it makes you feel any better, I think there's a lot of us still trying to figure that one out.

If you haven't noticed, while I believe in more standardization then a lot of manufacturing people, I'm not at the far end of the everything must be standardized spectrum. Here's some random comments based off my experience:

A. Not all standards need to be documented and there can be more than one "right" way to perform a task.

B. There is some good information / requirements that can be standardized not for the sake of controlling people's actions, but to serve as a knowledge reference to either benchmark against during corrective action or improvement activities or to let you know that something has changed in the system.

For instance, does it really matter if we set a standard to change a weld tip every 950 - 1050 welds and an operator changes one at 945 or 1055? But if an operator is changing at 850 or 1150, maybe we have a bad batch of tips, or maybe he has developed a method that will allow us to extend our tip life. The key is, that he lets us know, so we can ask "why", not because he did wrong and needs to be punished or retrained.

C. Getting operators to buy-in takes a long time if that's not your culture. As someone from a solid "world class" manufacturer told us the other day about an improvement he was attempting to, "I can get the guys on my shift to do it, but the other shifts is another story" It happens everywhere!

How to measure "overstandardization? How much resistance are you getting? While putting 90% of the salt on a piece of meat might not be enough, putting 110% might cause someone to get sick. So given the choice, are you better off with too much or too little.

To get people to buy into the overall system, you often need to wait for a chance when they have an issue with something and they can benefit from standardizing. Then when they find out that others don't buy in, they might buy in a little more next time to one that someone else proposes.

I once heard that when implementing a new methodology 20% will buy in right away (but sometimes this is not good, as they don't question or understand the changes well enough). 60% will sit on the fence, and you need to get some successes before you can get them to buy in. 20% will just be ornery and possibly never buy in. While you can't let the last 20% disrupt your goal, don't spend too much time focusing on them. If anything, leave them in the dirt with as you grow and hopefully someday down the road they will buy in.

Don't get discouraged. Sometimes you have to let them feel like they "got one over on you" before they buy in.

Keep trying and smiling.


Try to learn their "mental model" as to why they don't feel there isd a need to standardize or follow the specific standard. This can be a great learning opportunity.

D. Not all standards carry the same weight. Focus first on what's important and what you can get users to buy into.

Add new comment

More information
  • Files must be less than 2 MB.
  • Allowed file types: png gif jpg jpeg.
More information
  • Files must be less than 2 MB.
  • Allowed file types: zip rar.