ORIGINAL COMPANY-CONTEXT PRACTICE
Distinguish first-pass from final yield
Original practice scenario inspired by Texas Instruments Incorporated's public business context; this is not a reported interview question or a description of its internal systems. A synthetic manufacturing database stores device test attempts with attempt_number, outcome, and test_program. Explain how to compute first-pass yield and final observed yield without counting retests as new devices.
Written for independent practice using company context; this is not a verified question from a company interview.
A REASONED APPROACH
Model answer.
Use one device as the denominator and select attempts according to each metric's definition. First-pass yield describes the initial eligible attempt; final observed yield describes the latest eligible known result.
Work through it
- Define the device population, observation cutoff, and test-program comparability. Validate uniqueness of the ordering key for attempts on each device.
- Use ROW_NUMBER with ascending attempt order for first results and descending order for final results. Aggregate pass indicators over the same eligible device population.
- Report unresolved devices and the follow-up window. Test a fail-then-pass device, repeated failed attempts, duplicate ingestion, and an attempt arriving after cutoff.
Tradeoffs to discuss
- Final observed yield improves with additional follow-up, so recent and old cohorts may not be comparable. Program changes may require separate reporting strata.
Common pitfalls
- Counting passing attempt rows over all attempt rows measures attempt success, not device yield.
- Selecting MAX(outcome) invents an ordering unrelated to test chronology.