Key takeaway
Two correct percentages can describe different populations. Keep the master basis, phase inputs and quality decision visible.
A yield needs a basis and a quality context
A batch archive that retains only a final yield percentage cannot show what was compared. The denominator might be a master theoretical quantity, actual input to one phase or an intermediate quantity after earlier losses. Ask which measure the field represents before comparing batches or labeling a result abnormal. A production quantity and a quality-release decision are also different evidence states.
FDA's production/process Q&A explains that process-specific rejects can produce lower-than-theoretical yields and that the cause and consistency matter. For historical US finished-pharmaceutical context, the 2025 annual CFR edition describes actual and theoretical phase yields and master-record investigation bounds. Those exact 2025 passages were read 10 October 2026; the current eCFR yield text was inaccessible. Confirm the currently applicable rule and approved procedures before any real manufacturing or release action. This guide reconstructs evidence, not batch compliance.
Reproduce a two-phase example
This hypothetical batch has a master theoretical quantity of 100,000 units. Phase one produces 98,000 units, giving 98% against that master quantity. Phase two packages 95,000 units. Its local conversion is 95,000 / 98,000, or about 96.94%, while the cumulative packaged-to-master measure is 95,000 / 100,000, or 95%. The local conversion label here is an illustrative operational definition, not an assertion that it is the approved regulatory yield field.
Multiplying the unrounded phase conversions reproduces the cumulative 95% result. Averaging 98% and 96.94% gives approximately 97.47% and answers neither question. The phase-two tally contains 95,000 packaged units,1,500 documented sampling units,1,200 documented scrap units and 300 units not reconciled. It sums to 98,000, but the unexplained 300 must remain a separate category; assigning it to normal scrap would invent the missing explanation.
| Yield or reconciliation item | Illustrative result | What it establishes |
|---|---|---|
| Phase 1/master | 98,000 / 100,000 = 98% | Against stated master quantity |
| Phase 2/actual phase input | 95,000 / 98,000 ≈ 96.94% | Local conversion basis |
| Packaged/master | 95,000 / 100,000 = 95% | Cumulative quantity basis |
| Phase 2 accounting | 95,000 + 1,500 + 1,200 + 300 = 98,000 | Quantity tally complete |
| Unexplained category | 300 units | Cause and authorized resolution absent |
Store the master basis with the observed phase
Retain the relevant master-record version, product/batch identity, phase, theoretical quantity, actual input, actual output and units. Preserve the counting convention: gross intermediate output, conforming output and released quantity can differ. Link measurements, corrections and checker records to their source evidence rather than replacing them with a rounded dashboard percentage.
The 2025 master-record provision includes specified yield bounds tied to investigation. In an actual historical archive, retain the applicable approved range and which version applied at the time. Do not borrow a range from another product or reverse-engineer one from the distribution of successful batches. A repeated result can be consistent and still require review under its own approved basis. The source does not provide a recommended threshold for this fictional product.
Keep explanation and disposition separate
For the example, the arithmetic reviewer can reproduce the quantities and flag 300 units without an explanation. That completed result does not identify the cause or determine whether the batch is acceptable. Seek the actual investigation reference, findings, authorized decision and any evidence correction. If a later document reclassifies some units as sampling, preserve the original tally and the reason for changing it.
FDA's Q&A uses process-specific examples to explain why expected and excess rejects deserve different interpretation. It does not justify applying one process's normal-loss explanation to a different archive. Retain the local approved basis and evidence of what happened in this batch. A cleared quality status should link to the responsible decision rather than being inferred from a small percentage difference or a tally that happens to balance.
Choose what the archive can credibly support
The useful artifact is a phase-to-master reconciliation with formula definitions, observed quantities, range versions and unresolved investigation links. It can establish that cumulative yields are reproducible while some loss causes or quality decisions are missing. This differs from food-lot lineage, which follows inputs and outputs across lots, and inspection sampling, which reconstructs a lot-acceptance rule. Here the decision concerns yield basis and reconciliation across production phases.
Use Inventory and Due diligence to discuss evidence categories through synthetic rows first. Pharmaceutical records can expose formulas, product details, employee information and confidential manufacturing methods. VOID charges no upfront seller referral fee and may receive conditional receiving-program compensation. A named-recipient metadata introduction requires permission; actual samples, evaluation and licensing need separate review. Neither a yield figure nor the illustrative count supports a data-price or buyer-demand claim.