When a batch does not reconcile, the first reaction is often to recalculate the yield repeatedly in the hope of finding a transcription error. Occasionally that works. More often the cause lies in the process and not in the arithmetic.
Step 1: Confirm the failure is real before investigating anything
Before treating this as a deviation, rule out a documentation error.
- Re-check the BMR entries against the actual weighing/counting records, not against a summary sheet someone typed up afterward.
- Confirm that the theoretical yield you are comparing against is correct for this batch size. A copied figure from a different batch size is a common cause of a "failure" that is not a failure, and it is easy to prevent.
- Check whether all sub-lots or split batches were included in the total, or whether one container was recorded separately and forgotten.
If the numbers still do not tie out after this check, you have a genuine reconciliation failure and it is time to move stage by stage.
Percentage Yield Calculator, which compares actual with theoretical yield and shows the loss.
Step 2: Work backward from packing, not forward from granulation
Most people instinctively start at the beginning of the process and work forward. It feels logical, but it is usually slower. Working backward from the last stage tends to isolate the problem faster, because the closer you are to packing, the fewer variables are in play and the easier it is to get a clean count.
- Packing stage: verify reject and rework quantities, line clearance records and any partial cartons or blister strips set aside for QC sampling. Many "losses" turn out to be sitting in a QC retention sample or a rejected-carton bin that nobody added back to the total.
- Coating stage: check pan loading and unloading counts, and compare them with the samples pulled for in-process testing (weight gain, dissolution, hardness). These samples are legitimate, but they are often left out of the reconciliation.
- Compression stage: review the in-process rejects (tablets removed for weight, hardness or thickness out of range), start-up and shut-down rejects, and any granule left in the hopper at the end of the run.
- Granulation stage: check for material stuck to the blender walls, screen residue from milling, and moisture loss between wet granulation and drying. The moisture loss is expected, but it must be accounted for separately from real material loss.
At each stage, find out whether material was physically removed and, if it was, whether the removal was recorded. Most reconciliation gaps are not lost material. They are material that was never recorded.
Check each stage against its expected range as you go:
Step 3: Separate the loss into three categories
Once you have walked the stages, sort what you have found into three buckets.
| Category | Description | Typical example |
|---|---|---|
| Accountable loss | Removed on purpose and documented, or documentable after the fact | QC samples, in-process rejects, line clearance rejects |
| Process loss | Expected and within historical range, but not individually logged | Dust extraction, moisture loss on drying, equipment hold-up |
| Unexplained loss | Cannot be tied to any recorded activity or known process step | This is the only category that should drive a deviation/CAPA |
If everything you found falls in the first two categories and the total is now within the accepted reconciliation range, the investigation is closed. It was a documentation gap, not a material loss. If a real unexplained gap remains, it becomes a deviation.
Step 4: Check equipment hold-up specifically
Hold-up volume is worth its own step because it is the single most under-checked cause of a false reconciliation failure, and it is not the same for every machine. Blenders, mills, coating pans, and dedusters all retain a certain amount of product in ports, baffles, and transfer lines that does not show up unless someone physically checks.
- Compare with the equipment's known hold-up figure from qualification or from earlier batches of the same product and batch size. A value far from the history is a finding in itself.
- If hold-up has not been characterized for this equipment/product combination, this is a good moment to start recording it going forward so future investigations are faster.
Step 5: Interview operators before concluding
Numbers only tell part of the story. A five-minute talk with the operator who ran the stage often brings out something the paperwork missed, such as a spill that was cleaned up and reported verbally but never logged, a rework done off the record, or an extra sample pulled for a customer complaint. If you skip this step and rely only on records, the investigation tends to close with "cause could not be determined". That satisfies no one, and the same gap is likely to come back.
Step 6: Record the root cause, not only the finding
There is a difference between describing what happened and identifying why it happened. "Reconciliation was below limit due to unrecorded QC sample withdrawal at coating stage" is a finding. The root cause goes one level deeper and asks why the withdrawal was not recorded. The cause may be a missing SOP step, unclear responsibility, or a form without a field for the entry. The CAPA should fix that gap, not only correct this one batch's paperwork.
The loss mechanisms behind each stage are described in Understanding Yield Loss Points Across Granulation, Compression, Coating, and Packing.