Error Prevention versus Training

Six Sigma – iSixSigma Forums Old Forums General Error Prevention versus Training

Viewing 14 posts - 1 through 14 (of 14 total)
  • Author
  • #33961

    Ed Dignan

    I’m looking for thoughts on the benefits of building error proofing into the developmnet a complex order entry system, versus building and delivering training to “teach people how to do it right ” after rollout.
    From the data collected during testing, I belive I can justify the costs of building error proofing into the software,  against the benefit of reduced cost of orrors. But the organization wants to roll fast, and is accustomed to releasing on time, then multiple versions and repeated training to improve. Any other opinions or tool ideasbefore I make my case?  


    Ralph Gentile

    Sounds like your problem is not whether you can improve the system, but selling the change in behaviors.  Has your company taken any “hits” from customers over errors?  Are the process owners customer-centric?  How high up do you have to go to sell this?  How customer-centric is your division / company?
    Whatever your approach, focus on the benefits for all stakeholders.  Consider the response “we’ve always done it this way”.  Find out the dynamics that cause fear of change (it’s often that the powers that be fear lack of control) and have a defensible plan for “babysitting” the transition to the new behaviors.
    Can you pilot a small segment to success to create demand for your solution?



    Hi Ed,
    Most of my experience is with high volume complex manufacturing environments with complex order entry systems, and the largest Pareto reason on the 1st level Pareto for Returned Goods or Credit Memos has usually been Order Entry errors. My own two cents: I think you are on the right track assuming you have good data and have down your homework- gone to the source-removed bias- asked open ended questions, good well defined problem statement which is tied into the business objectives and financial impact (cost avoidance) well defined. etc. A good approach is always to work toward defect prevention rather than detection. Mistakes are not inevitable, where there is a process, there is always some creative and highly talented individual who can create an error.
    A couple of cautions here though:
    Never go to one solution early in the process. You can become emotionally attached to that one solution because you committed to early to it before all the factors to make a decision have been addressed. Force yourself to have at least 3 alternatives (3 to 6 is best) even if some of these seem like they are a shot in the dark. If you find yourself defending one of the solutions early on, -you have become attached. Get cross a functional team to look at prioritized solutions. Management has a tendency to listen to a Team more than to an individual and for good reason
    If allowed, soft fixes will be used the majority of the time because they require little effort. They usually go something like this: Error-Lack of training-Countermeasure-Train the operator who made the error. Or: Error-Operating Standard not clear-Countermeasure-Add more info to the standard and retrain.’ These still produce defects.
    If you really want to improve the process the best results are usually obtained by taking the high but hard road. I’m not saying throwing money at the problem instead of intellectual capital, I’m saying build Quality into the process, while keeping in mind the proper payback analysis rules.
    Good luck and regards,



    I come manufacturing and assembly back ground. Though training of people definitely help in effectively containing the errors but not sure of 100% of times, mainely due to montony and repeated operations, and some times they commit and pass the errors. It is better to have the error proofing stuff in place rather that trying to contain / correct it later.
    With wishes



    Poka Yoke systems are imperative in any process!  An order entry system is similar to a financial system I helped re-design. Normally you have high turnover in these posisiton and training is only slightly better than OJT.
    Insist on the Poka Yoke and show the cost benefit in customer sat and returns when entered wrong.



    You said that you are developing a “complex order entry system.”  That is your first problem.  Order entry systems can be as complex as necessary at the back end, but the user interface needs to be simple to use and include tools that enables the users to do their job.  If that is not the case, “error proofing” alone won’t solve your problems.
    If it is too late to redesign the user interface, in addition to using various Six Sigma tools to improve performance, develop job aids for the users and provide hands-on training that includes a training database and exercises that challenge the trainees to solve real-life problems using the order entry system. 
    You cannot expect order entry clerks to navigate a complex system in an accurate and timely fashion without training.  If that training is online computer or web based training, new users will be able to take the training at any time (without waiting for a class to form) and will be able to refer to the training as a resource tool.


    Debra Mallette

    For the business case for error prevention and training, the focus needs to be on the sustaining costs for the system.  The return is customer satisfaction.  Consider the sustaining costs determined by the process error rate and variance.  Prevention is a one-time cost that establishes a lower average error rate and a lower sustaining cost base rate with a higher impact on customer satisfaction. Training is an ongoing fixed sustaining cost. Training’s primary effect on the error rate is to reduce the error rate variance so it’s affect on customer satisfaction is less direct.  The question is whether you and your organization have a handle on these projected sustaining costs and have used them as part of the business case – particularly since the prevention costs will come out of the project budget and the sustaining costs tend to be part of the fixed cost for operations and will coming out of someone else’s budget.  This is why a previous poster’s message of including all stakeholders in critical. I have had my operations and finance stakeholders successfully advocate a change to the schedule so that the sustaining cost is measurably lower.



    As someone mentioned earlier, no point trying to error proof a real complex system. You really need to simplify. Some Lean tools would help here. Why don’t your customers go straight to the point of activity and order from Production directly using pull systems. Order entry only has to happen once to set up the customer. Simpler systems will always work better no matter how much fool proofing you do. Good luck. 


    Vlad Vasilescu

    The premise of my response is that:

    no system is perfect,
    application fail proofing (AFP) for all errors may be cost prohibitive,
    it is possible to sort out what shall be AFP,
    it is possible to build redundancy in the human sub-system.
    Your System is characterized by an application-human interface. The interface and the error and fault propagation analysis will tell you:

    what errors are at stake: character or logic,
    where they occur,
    how much they costs “as is” or “rework”
    where in the propagation path it is practical to institute controls and “rework”,
    sort out what to AFP or build into the human sub-system controls.
    You are talking about training, but actually your human “redundant” sub-system shall be proceduralized. The procedure may only address fault detection and correction.
    To speed up the learning process you could:

    simulate error propagation in meetings with the users,
    create competition between or award users for error detection and correction. You have to build a sense of “pride” that the humans are smarter than the application.
    If you do some of the above simulations before conversion, the customer will be less exposed.
    If you can’t, monitor cost of errors in post conversion. As time is of essence, you could have later a version with more AFP.
    Good luck, Vlad



    I don’t undersand why you would even want to support a complex order system.
    Someone orders a part, you ship it to them.  Invoice goes to accounting for billing, send a bill. Payment come in, close billing.   Kanban goes to manufacturing, manufacturing builds what you sold.  Parts you use to build, generates a Kanban for replacement to your supplier.  And the process starts all over again.  Finished goods, and component parts are at Min/Max.  Inventory at Min/Max, minimum lot size. (Visual control)
    A grocery store carries hundreds and hundreds of different items.  You go to the store, pick what you want (order), go to cashier for cost (billing) pay bill, take goods.  Store reorders what everyone purchased for the day.  Lag is 2 days, so they carry 2 days stock.  Shelf below min, restock.


    Ed Dignan

    Thanks for all the feedback on this question. Good to see consensus around the need to justify and build in error proofing rather than rely on training to reduce errors. A clarification – I wrote “complex order entry system” in order to keep the original message brief. It is a complex system in that  it’s the entry  system for custom printed products: it handles about 300,000 custom orders per year, entered from service reps in 40 locations. For each order, it assembles up to 84 print  features, prices the product, matches the order file to a graphic file ( up to 131 graphic features )  to ensure they match, and routes the order file and the graphihc file to any of 14 production plants depending on product / feature  specifics. The specification assembly application  is mature and works well;  the current project is to replace the graphics file application.


    Martin Schmalenbach

    I’m a training manager in a manufacturing organisation that is world #1 in what it does in terms of volumes & turnover. We are about 18 months in to the journey that is Six Sigma. We had a similar issue with regards to operators following build and work instructions and the cost of quality was literally killing us. Pre six sigma we would have retrained the workforce.
    What we did was a simple DMAIC process and determined that there were actually 2 linked improvement actions
    – first, reformat and clarify the work instructions to the extent that they are not open to interpretation (making them in the process much bigger)
    – second, to strengthen the training process to reduce variation by ensuring all new people were trained to the same (more clearly defined) standards by trained trainers who are all themselves trained the same way.
    The control step involved new policies around operators having to physically follow the work instructions.
    We didn’t retrain anybody, thereby avoiding the implied insult to the operators, and we have a credible, robust training system. Revised build instructions alone would not have produced the improvements we are seeing, any more than just training some trainers would have.
    We have piloted this and are now rolling it out across all parts of the business. Bottom line improvements are heading towards a sustained 4 million annually, returning the company to profitability for a 1-off spend of 40K on training and 250K on labour to reformat the many build instructions.
    As a result of this project a number of other issues were identified and poka yoke and wider DFSS activities are now being implemented.


    John G.

    It appears that the first boulder in your path to error proofing is by in from your organization.  The previous replies have some excellent advice on how to break through that problem.
    Once you overcome this problem, then I suggest going back to basics and conducting a Design FMEA.  Use a cross-functional team to determine what can fail and then prioritize how to eliminate the problems.  I also agree with a previous response, focus on prevention, not on detection.
    Best of Luck.


    Rosalie Hopkins

    How soon do you need the data? In April, we will begin to roll out a project that might supply what you are looking for.

Viewing 14 posts - 1 through 14 (of 14 total)

The forum ‘General’ is closed to new topics and replies.