Rolled Throughput Yield measures the probability that a unit passes through every process step without defect, rework, repair, or scrap.
Definition
Rolled Throughput Yield (RTY) is the combined yield of a multi-step process. It estimates the chance that one unit can pass through all required steps correctly the first time. RTY is usually calculated by multiplying the first-pass yield of each process step.
RTY is stricter than final yield because final yield can hide rework, repair, sorting, and repeated attempts that consume capacity and delay flow.
History
RTY became common in Six Sigma and process improvement because organizations often reported acceptable final output while internal defect generation remained high. It helped expose the hidden factory: the extra work required to fix, inspect, and recover from defects before shipment.
When to Use
Use RTY when a process has multiple sequential steps and the team needs to understand total first-pass performance. It is valuable in manufacturing, healthcare, software releases, service workflows, claims processing, and transactional work where rework is common.
Step-by-Step
- Define process boundaries and major steps.
- Define what counts as a first-pass success at each step.
- Collect units processed and units passing without rework at each step.
- Calculate first-pass yield for each step.
- Multiply step yields to calculate RTY.
- Compare RTY with final yield to reveal hidden rework.
- Prioritize steps with the largest yield losses.
- Track RTY after corrective actions and process controls.
Examples
- Assembly: Five steps each look strong individually, but combined RTY shows only 72 percent pass all steps first time.
- Healthcare: A lab order must pass registration, collection, labeling, testing, and reporting without correction.
- Service: A customer application passes every review stage without return or clarification.
Common Pitfalls
- Using final yield and calling it RTY.
- Ignoring rework loops that happen before final inspection.
- Inconsistent definitions of pass and fail by step.
- Combining dissimilar products or work types.
- Not validating data capture at each step.
- Improving the worst step while ignoring system interactions.