A contemporary approach to problems with widgets is to:

• Look for obvious problems either from extensive experience, or a qualitative examination of the widget (assembled or disassembled).

• Quantitatively investigate through comparison with engineering or performance specifications.

There are two simple but powerful techniques that may be considered at all stages from prototype to pilot run and then to production to help understand why your widget won't work. These are:

• Components search and
• Paired comparison

This technique is applicable when:

• Your widget is capable of disassembly and reassembly;
• A performance output can be measurable and is repeatable;
• You may only need one  typical "good" and "bad" widget each. This methodology requires just two units in which to collect valuable information.

The basis procedure involves:

• Select the best and worst performing widgets.
• Based on engineering knowledge of the design of the widget, rank the widget components in descending order of importance.
• Switch the top ranked component from the "good" unit with that component in the "bad" unit.
• If there is no change, ie the "good" unit stays "good" and the "bad" unit stays "bad", then this top ranked component is unimportant. Go to the next ranked component.
• If there is a partial change of the out of control limit for one or both of the two units output, then this component is not the only important variable. Further interaction with other components can not be ruled out. Go to the next ranked component.
• If there is a complete reversal in output of the two units, then there is no further need to continue. The most important variable has been found.

This technique can be used when:

• The unit can not be disassembled;
• A suitable parameter must be found to distinguish "good" from "bad" and
• Usually need four to eight pairs of "good" and "bad".

The steps involved are:

• Select one "good" widget and one "bad" widget, drawn close to the same production time.
• Observe in detail to note differences between the two units. The differences can be visual, dimensional, electrical, mechanical, chemical, etc.
• Select another pair of "good" and "bad" widgets. Observe and note the differences as above.
• Repeat the process with a third, fourth, fifth, etc pair until the observed differences show a pattern of repeatability. Consistent differences can be narrowed down to a few factors providing a strong clue to the significant cause(s).