Definition
Percentage of non-release tests, including stability pulls, method validation, and proficiency testing, completed by their original scheduled date rather than a rescheduled one. Detects scheduling algorithms that trade soft deadlines for release throughput.
The Measurement Gap
AI lab scheduling algorithms optimize for throughput and release testing cycle time. The metrics with the most immediate deadline pressure. Non-release testing (stability pulls, method validation, proficiency testing) has softer deadlines, so the algorithm systematically trades small delays on non-release testing for throughput gains on release testing. Each individual trade-off looks rational: push a stability pull back three days to free an HPLC for two release tests today. But the algorithm makes this trade-off every day, and the stability program absorbs the cumulative delay. Lab throughput metrics show improvement. Instrument utilization is up. Release testing cycle time is down. Every metric on the dashboard is green. Meanwhile, the stability program (the long-term product quality monitoring that determines whether shelf-life claims are valid) quietly falls behind schedule with no metric connecting the cause to the scheduling algorithm.
Decision Supported:
Is the AI lab scheduler systematically deprioritizing long-term testing programs in favor of short-term release testing throughput, and how far has the compliance gap grown?
Formula
Notation:
NRTSC = (Σ On_Time_Non_Release_Tests / Σ Scheduled_Non_Release_Tests) × 100
Variables:
- On_Time_Non_Release_Tests = non-release tests completed by the ORIGINAL scheduled date (not rescheduled dates)
- Scheduled_Non_Release_Tests = all non-release testing activities on the schedule for the period
Worked Example: Last month: 147 non-release activities scheduled. Stability pulls: 68 scheduled, 56 on time (82.4%). Method validation: 42 scheduled, 38 on time (90.5%). Proficiency testing: 12 scheduled, 12 on time (100%). Environmental monitoring: 25 scheduled, 23 on time (92.0%). Overall NRTSC = 129/147 = 87.8%. Pre-AI baseline: 96.2%. The 8.4-point decline is attributable to the scheduling algorithm's priority logic.
Edge Cases:
- "On time" means the original scheduled date, not a rescheduled date. If the algorithm pushes a stability pull from Day 180 to Day 185 and the test happens on Day 185, that is late.
- Stratify by test type: stability is the most critical non-release test. If NRTSC is low because stability is consistently deprioritized, that is worse than proficiency testing running late.
Sign convention: Higher NRTSC is better. Declining trend post-AI adoption indicates systematic deprioritization.
Proposed formula, pending validation pilots.