By Phillip Mueller and Umesh Saxena
Executive summaryThrough experience, groups of people within an organization learn what is called tribal knowledge. While each participant knows some things, the group fails to use its combined knowledge in the most optimal manner, and such knowledge is not transferred efficiently to other group members. Thus, tribal knowledge, common to all organizations, can be a source of process variation and inefficiency. Determining the impact of tribal knowledge on an industrial process and creating a method for improving the process can eliminate mistakes and improve efficiency.
A manufacturing firm currently uses a process based on tribal knowledge for planning customer specified products. The planning process specifies the manufacturing method, manufacturing sequence of events, tools, materials and manufacturing documentation for production of parts that are designed by the customer. The company’s management believes tribal knowledge causes high levels of variability within the planning process and that a standardized process would reduce variation, lessen quality issues, shorten production lead times and help train new operators faster.
This research project investigated and developed a standardized process to reduce or eliminate process variation caused by the use of tribal knowledge. The methodology contained three parts. First, conduct an experiment to determine if tribal knowledge affects the planning process. Second, analyze the current planning process to determine the specific process points that use tribal knowledge. Finally, create a standardized process to minimize or eliminate the use of tribal knowledge in the planning process.
The products made are liner-mounted individual parts that may be printed or unprinted. All parts are processed in roll form. Rolls of finished material may be broken down further and sold per piece, per roll or per sheet. The planning group consists of 10 full-time planners. The responsibility of the planners is to determine the production processes and create production documentation for parts that are designed by the customer. Parts designed internally by the company follow a different planning process.
Planners will specialize in certain types of products or production processes, but many are trained to plan a variety of part types. None are capable of planning all the product types made by the company.
In a typical year, the group will plan approximately 3,000 customerdesigned parts. Most of these parts, approximately 70 percent, never have been made by the company before and probably never will be made again. Planners determine the types of equipment used, production process sequence, and inspection required to make a part, along with all necessary production documentation.
Planners are required to choose the type of equipment to be used based on the requirements of the part, capabilities of the equipment and cost of production. Equipment capacity is not used to select equipment. New planners are trained by job shadowing.
New planners (trainees) observe as experienced planners (trainers) create new part plans. At the discretion of the trainer, the trainee will create part plans under the supervision of the trainer. As the trainee gains experience, the trainer will allow the trainee to create part plans without supervision. Beyond the trainer/trainee relationship, there is no structured method of planner training and no definition of a fully trained planner. Trainees with no experience can take from six months to one year before the trainer will consider them fully trained. After training is complete, it’s not unusual for newer planners to seek advice from more experienced planners when planning a difficult or new type of part.
The current planning process has issues with quality, training and incorrect part costing. Company officials believe that standardizing the planning process will improve these issues.
Currently, the planning process has an average monthly defect rate of 4,500 defective parts per million (dppm). Historically, the monthly defect rate has ranged between 250 dppm and 7,500 dppm. This number only includes product that has been shipped to the customer and returned as defective. Defects are directly attributable errors made during the planning process.
Training new planners is time consuming and costly. The six-month to 12-month process is the longest training period for any production level position. On average, a new planner enters training every six months.
Part costing is used to calculate profit margins, determine pricing and monitor productivity. Part costing is solely determined by part planning. There is no data to identify the impact of incorrect part costing; however, incorrect part costing happens routinely, evidenced by the number of parts that are re-planned. Re-planning only occurs when an existing part plan is used for a different but similar part.
To determine the impact of tribal knowledge on the planning process, each of the four test subjects was asked to plan the same two parts. The part plans from each test subject were compared to each other to determine the number of inconsistencies in the plans. After comparing the plans, the test subjects were interviewed in a team meeting to determine the reason for the inconsistencies. Each inconsistency was categorized by the team to determine which inconsistencies were due to tribal knowledge.
Once the tribal knowledge inconsistencies were identified, the planning process was analyzed to determine the process points where tribal knowledge inconsistencies were created. Standard process guidelines were developed for each process point to reduce the impact of tribal knowledge on the planning process.
To determine the impact of the standard process guidelines, a second experiment was conducted. Similar to the first experiment, two different test parts were planned using the standard process guidelines as part of the planning process. Comparing the results of the first and second experiments would indicate the impact, if any, of the standard process guidelines.
The four different parts selected for the experiments are typical of the types of parts that would be planned on a regular basis. For simplicity, the parts are referred to as A, B, C and D. Parts A and B were used in the first experiment. Parts C and D were used in the second experiment.
Four test subjects were chosen from the 10 planners currently in the planning department. To take part in the experiment, each test subject had to meet three criteria. All of the planners that met the criteria took part in the experiment.
First, each participant had to be trained fully in the planning process. Untrained or partially trained planners do not plan parts on their own, so their participation in the experiment would not be an accurate representation of the current planning process.
Second, each participant had to be engaged actively in the planning process. Former planners or those who plan production on an intermittent basis always work with a trained planner, so their participation would not be an accurate representation of the current process.
The last requirement was that at least one participant should be drawn from each of the three process groups. Drawing participants from each of the three groups ensures a variety of planning experience and accurately represents the current process, since planners often plan orders for processes outside of their primary planning area.
For every part that is produced, the planners create a part plan. Each part plan consists of the following five documents:
Each document contains several pieces of information. Each piece of information is a document element. For each document type, critical document elements were identified. Critical document elements are defined as pieces of planning information that are critical to accurate costing and production of the part. The project team determined which document elements are considered critical. For this experiment, only critical document elements were evaluated for inconsistencies.
Each of the five documents from the four test subjects was compared to the other to determine the number of critical document elements with inconsistencies. An inconsistency is any critical document element that differs between two or more test subjects.
When comparing the part plans from the four test subjects, a total of 30 inconsistencies were identified out of a possible 57 critical document elements. Inconsistencies were found in all five of the document types.
The project team held a meeting to categorize each inconsistency identified in the first experiment by its type and its impact on the production process. Categories were assigned based on the project team’s knowledge of the planning and production processes. Three types of inconsistencies — tribal knowledge, lack of standardization and error by planner — were identified. The types indicate the source of the inconsistency in the planning process.
For the first inconsistency type, tribal knowledge, planners made different planning decisions based on their individual understandings of the production process. All of the planners thought they were making the correct decision based on their understanding of the production process. For example, one planner required a print quality test for all printed products, but other planners required the print quality test only for particular types of printed products. All of the planners could support their decisions based on different understandings of the production process.
For lack of standardization inconsistencies, planners made different planning decisions because they were not given guidance on how to make the decision. Most inconsistencies were administrative in nature. For example, no standard naming convention has been established for new parts, so planners created their own method. Lack of standardization inconsistencies are not related to a planner’s understanding of the production process and rarely have a significant impact on the process. These inconsistencies can be eliminated by creating a standard for the critical element and communicating it to the planning group.
For the third inconsistency type, error by planner, planners admitted to making an error and understood why an incorrect decision had been made.
Each inconsistency was assigned one of four impact levels. The impact levels indicate the worst-case consequence of each inconsistency identified. Three of the impact levels are considered serious and require immediate attention. The first serious impact level is “inefficient production process.” An inefficient production process inconsistency could cause a less-than-optimal process to be used for production. The second serious impact level is “product defect.” A product defect inconsistency could cause a defective product to be produced. The last serious impact level is “incorrect or variable costing.” An incorrect or variable costing inconsistency could cause the ERP system to calculate the production cost incorrectly.
The only impact level not considered serious is “minimal impact.” A minimal impact inconsistency would not cause a noticeable negative impact to production or the customer. This type of impact does not require corrective action.
Tribal knowledge inconsistencies that had an impact level other than minimal were the focus of this research. All other inconsistencies can be addressed via other activities or, in the case of minimal impact inconsistencies, can be disregarded.
The project team identified decision points in the planning process at which tribal knowledge inconsistencies were created. First, the team mapped the planning process and identified what decisions were made at each process step. Then, each tribal knowledge inconsistency was assigned to the process step where the inconsistency was created. In some cases, inconsistencies were created due to decisions made at more than one process step.
Last, the team identified the type of decision that was being made at the point where it created the inconsistency. The result was the identification of four decision points where tribal knowledge inconsistencies were created. The decision points are:
Six inconsistencies occurred at the point where the planner decides which machine(s) will be used to make the product.
Five inconsistencies occurred at the point where the planner decides which tool(s) will be used to make the product.
One inconsistency occurred at the point where the planner decides which ink(s) will be used to make the product.
Four inconsistencies occurred at the point where the planner decides what information and checks will be included on the quality plan.
Figure 1 is the map of the planning process with decision points indicated. Figure 2 shows a sample of the types and impacts of each of the inconsistencies identified in the experiment, along with the decision points at which the inconsistencies were created. Standard process guidelines were created for each of the four decision points based on planner experience, process standards and material standards. The decision making guidelines will replace tribal knowledge by walking the planners through a scripted process at each of the four decision points. The intent of the guidelines was to standardize decision making.
The four test subjects were trained on the standard process guidelines. As a team, the test subjects walked through each of the guidelines line by line. Team members were encouraged to ask questions and challenge the guidelines when they did not agree with or understand the guidelines. At the end of the training, all the test subjects agreed to use the guidelines when planning future orders.
The second experiment, which determined the impact of the standard guidelines on the planning process, was conducted in the same manner as the first experiment, except for the use of different parts. Parts C and D, which were similar to but not exactly the same as parts A and B, were used in the second experiment. Using different but similar experimental parts gave the test subjects the opportunity to create new part plans using the standard process guidelines without significantly changing the difficulty level of the experiment. Data from the second experiment was analyzed and categorized the same way as the data from the first experiment.
The implementation of standard process guidelines reduced the number of inconsistencies related to tribal knowledge in the planning process. The second experiment produced 18 critical document element inconsistencies. Of the 18 critical document inconsistencies, two were categorized as tribal knowledge inconsistencies. So in experiment No. 2, 11 percent of the inconsistencies found were tribal knowledge inconsistencies. In experiment No. 1, 40 percent of the inconsistencies found resulted from tribal knowledge inconsistencies. Figure 3 shows the inconsistencies found in the second experiment, along with their categorization.
This study found that tribal knowledge is a significant source of inconsistencies in the planning process and that standardizing decision making, in this case through the use of process guidelines, can reduce the number of inconsistencies. However, the use of tribal knowledge in decision making was not the only source of inconsistencies. Nevertheless, the introduction of standard guidelines reduced the introduction of inconsistencies that could cost the company time and money.
A review of the literature shows limited research on the topic of tribal knowledge and its impact on organizations. An additional area of research could be how organizations communicate such knowledge. Understanding how informal knowledge is communicated is important for understanding how an organization uses tribal knowledge.
Quantifying the value of individual pieces of tribal knowledge is another area with little research. Not all tribal knowledge is important, and figuring out which pieces are relevant could help organizations identify what information needs to be distributed throughout the organization.
Phillip Mueller is enrolled in the industrial engineering Ph.D. program at the University of Wisconsin-Milwaukee. He has 20 years of industry experience in a variety of engineering and business roles, including quality management, international project management, and deployment of Six Sigma problem solving and lean methods in both manufacturing and transactional businesses. He earned his M.S. in industrial engineering from the University of Wisconsin-Milwaukee and his MBA from Marquette University.
Umesh Saxena retired from the Department of Industrial and Manufacturing Engineering at the University of Wisconsin-Milwaukee in 2008 after 40 years there. He taught a variety of courses at the graduate and undergraduate levels, including engineering economy, quality control, statistics and facilities planning. He obtained his B.S. in mechanical engineering from the University of Roorkee, India, and his M.S. and Ph.D. degrees from the University of Wisconsin-Madison. He has published research articles in The Engineering Economist, Energy Engineering and IIE Transactions. He served as the director of the Industrial Assessment Center from 1986-2006. He was elected a fellow of IIE in 1994.