Quality & DefectsIntermediate

Root Cause Analysis for Garment Defects: A Practical Method

Repairing defects fixes garments; root cause analysis fixes the process. A step-by-step method for garment factories using defect data, the 5M categories, fishbone diagrams and 5 Why, with a worked example on open seams.

By Published 3 min read
On this page

Root cause analysis (RCA) is the work of finding why a problem happens, so that the fix stops it from happening again. In garment factories, the usual response to a defect is to repair the garment and remind the operator. That treats the symptom. RCA looks for the cause in the machine, material, method, operator or environment, and changes the process.

Step 1: Define the problem precisely#

Write a problem statement that says what, where, when and how much:

Open seams at the left side seam of style 4521, found at end-of-line checking on line 6, 18 per 400 pieces on Tuesday, up from 3 per 400 the previous week.

A vague statement ("quality is bad") cannot be analysed.

Step 2: Collect facts#

  • Defect counts by operation, operator, hour, machine, fabric lot and size.
  • Photos of the defect.
  • What changed recently: new fabric lot, new thread, machine repair, new operator, new style.
  • Observation of the operation as it is actually done.

Data often narrows the problem quickly. If all open seams come from one machine, the search is much smaller.

Step 3: List possible causes (fishbone)#

A fishbone (Ishikawa) diagram groups possible causes into categories. Garment factories often use 5M plus environment:

Category Examples for open seams
Machine Wrong stitch settings, worn feed dog, knife cutting too deep, tension too loose
Material Slippery fabric, fraying edges, weak thread
Method Seam allowance too narrow in the pattern, wrong guide setting
Man (operator) Not aligning edges, pulling the fabric, new operator
Measurement Checker not stretching seams to test, unclear standard
Environment Poor lighting at the station

Step 4: Find the root cause#

For the most likely causes, ask "why" until you reach a cause you can act on and control. See 5 Why analysis.

Then verify: test the cause. Change one thing and see if the defect stops, or reproduce the defect by creating the suspected condition.

Step 5: Act#

Decide on:

  • Containment: protect the customer now (sort and repair affected garments, check bundles in WIP).
  • Corrective action: remove the root cause (reset the machine, change the pattern, retrain with a standard).
  • Preventive action: stop it happening elsewhere (add to maintenance checklist, update standards for similar styles).

Record these in a CAPA form.

Step 6: Check the result#

Measure the defect rate after the action for long enough to be sure. If it has not dropped, the root cause was not found; go back to step 3.

Worked example#

Step Finding
Problem 18 open seams per 400 at left side seam, line 6
Data 16 of 18 from one overlock machine; started Monday afternoon
Change Machine had a knife replaced Monday morning
Observation Knife cutting 2 mm more than the seam allowance, so stitches catch less fabric at curves
Root cause Knife position not reset to standard after replacement; no setting check in the repair procedure
Containment 100% check of side seams in WIP and finished stock of the style
Corrective Knife reset to standard; seam width verified
Preventive Setting check added to the knife replacement procedure for all overlock machines
Result Open seams back to 2 to 3 per 400 over the next week

Common mistakes#

  • Stopping at "operator mistake". Ask why the mistake was possible.
  • Acting on the first idea without checking data.
  • Several changes at once, so you do not know which one worked.
  • No follow-up measurement.

Use the root cause analysis sheet to record each case.

  • DHU: finding which defect to work on first.
  • Sewing defects: typical defects and their causes.
  • Inline inspection: where many root cause investigations start.
  • Kaizen: building the fix into continuous improvement.

Was this useful?

Related defects

Related technical notes