ORIGINAL COMPANY-CONTEXT PRACTICE
Detect calibration drift without confusing a product-mix change
Original practice scenario: A hypothetical equipment analytics team sees inspection measurements shift after a new product mix enters production. How would you investigate calibration drift?
Written for independent practice using company context; this is not a verified question from a company interview.
A REASONED APPROACH
Model answer.
Separate measurement-system changes from changes in what is being measured. An aggregate shift alone cannot identify whether calibration, process conditions or product composition caused it.
Work through it
- Retain instrument and calibration versions, product identifiers, reference measurements and environmental covariates at the observation level.
- Compare stable reference samples and like-for-like product slices over time, using a predefined drift statistic and uncertainty rather than searching for the most alarming chart.
- Examine missing data and maintenance events, then validate a suspected shift with an independent reference or controlled check before changing downstream correction factors.
Tradeoffs to discuss
- Reference checks improve attribution but consume measurement capacity; observational monitoring offers wider coverage while leaving more confounding.
Common pitfalls
- Do not normalize away a genuine process change merely to restore a familiar average.
- Changing the reference population or instrument software requires a baseline version so monitoring remains interpretable.