The Key to Proper Root Cause Identification

Step 1: Ensure the root cause is identified as soon as an issue is discovered

Frequently, when a part is identified as scrap or rework, the operator knows exactly what caused the issue. (Perhaps they just put in an incorrect offset, or they didn't fixture the part properly.) Unfortunately, many companies simply have these operators tag the parts and put them in an MRB cage, and only days later does a quality engineer begin writing up a non-conformance report and return to the operator to ask them what went wrong with the part. That operator has since run multiple parts and jobs, and frequently cannot recall exactly what went wrong.

A better approach is to capture at least a suspected root cause immediately after the issue is discovered. When using Net-Inspect, as soon as an out-of-tolerance measurement is entered, an operator is prompted to specify a root cause and corrective action. Individuals such as quality leads or engineers are notified via email and directed to the issue. This allows them to arrive within minutes of the non-conformance occurring and validate that the specified corrective action was accurate.

Step 2: Ensure that root cause options are appropriately granular

Recently, we visited a company with nearly $1M of annual scrap and rework. They were entering non-conformance information in their ERP system as soon as each issue occurred, but they only made six root causes available to their operators and inspectors:

1. Casting issue
2. Damaged
3. Design issue
4. Heat treat issue
5. Program
6. Other

We were not surprised to find that more than 60% of their scrap and rework volume had been assigned to the "damaged" and "other" categories. The quality manager believed that offset errors were the most costly and frequent cause, but because of the absence of data to back up that theory, no action could be taken. Instead, they were selecting 1-2 of the most expensive non-conformances within a week and doing an 8D or 4D analysis the following week. Not only was this occurring far too late, it was also incredibly time-consuming. And with such a small sample size, hardly any actionable insights came from the exercise.

The key to proper root cause identification is to strike the right balance: not so granular that you have a unique root cause for every scrapped part, but also not so broad that everything is assigned to a generic category such as "Labor" or "other".

Step 3: Ensure that root causes are, in fact, root causes

On non-conformance reports, we sometimes see "Setup Piece" listed as a root cause. However, a setup piece cannot itself be responsible for features being out of tolerance, so this is not a true root cause. "Setup Piece" is just a description of what was impacted; selecting this as a cause only obscures the actual root cause, preventing you from determining why a setup piece is needed in the first place.

What you want to see on your non-conformance report is an exact explanation of what happened. (Was the CNC program faulty? Was the wrong tool used?) When this level of detail is provided and available for analysis, teams can leverage any new insights and use them to incorporate attribute checkoffs into their inspection plans. This way, the next time the part number is run, the operator will be reminded of what needs to be done to ensure that parts are not consumed in setup.

Step 4: Leverage reporting tools to eliminate recurrence of causes

Once you have begun identifying root causes with the appropriate level of granularity as soon as incidents occur, a wealth of reporting and insights will become available.

One of our most-used reports inside of Net-Inspect is a "Root Cause by Cost" report. It not only allows companies to determine which root causes are occurring most frequently, but also which root causes are costing the company the most.

Companies often use this report to get buy-in for the software or equipment needed to fully eliminate a cause. For example, the report may reveal that the total scrap caused by tool breakage is far more expensive than a proper tool life management system would cost, or that it would be cheaper to calibrate a particular machine than to continue allowing it to produce parts that will later be scrapped.