A reconciliation failure is one of those events that looks simple on paper and turns messy the moment you start digging. The number doesn't tie out, someone flags it, and now there's pressure to explain it fast. The temptation is to jump straight to recalculating the yield percentage over and over, hoping a transcription error will jump out. Sometimes it does. Most of the time it doesn't, and the real cause is sitting somewhere in the process, not in the arithmetic.
This isn't a formula walkthrough — you already have a calculator for that. This is the investigation itself: where to look, in what order, and how to avoid the two most common mistakes people make when a batch doesn't reconcile — chasing the wrong stage first, and closing the investigation before the root cause is actually confirmed.
Step 1: Confirm the failure is real before investigating anything
Before treating this as a deviation, rule out a documentation error. This sounds obvious, but it's the fastest possible resolution and skipping it wastes everyone's time.
- Re-check the BMR entries against the actual weighing/counting records, not against a summary sheet someone typed up afterward.
- Confirm the theoretical yield used for comparison is correct for this batch size — a copy-pasted theoretical figure from a different batch size is a common and entirely preventable cause of a "failure" that isn't one.
- 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 don't tie out after this check, you have a genuine reconciliation failure and it's time to move stage by stage.
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's 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/rework quantities, line clearance records, and any partial cartons or blister strips set aside for QC sampling. A surprising number of "losses" are sitting in a QC retention sample or a rejected-carton bin that nobody added back into the total.
- Coating stage — check pan loading and unloading counts, and cross-reference against sample pulls taken for in-process testing (weight gain checks, dissolution, hardness). These withdrawals are legitimate but are frequently left out of the reconciliation math.
- Compression stage — review in-process rejects (weight/hardness/thickness out-of-range tablets removed during the run), start-up and shut-down rejects, and any granule left in the hopper at the end of the run.
- Granulation stage — check for material adhering to blender walls, screen residue from milling, and moisture loss between wet granulation and drying, which is expected but must be accounted for separately from an actual material loss.
At each stage, ask the same question: was material physically removed, and if so, was it recorded anywhere? Most reconciliation gaps are not lost material — they're unrecorded material.
Check each stage against its expected range as you go:
Step 3: Separate the loss into three categories
Once you've walked the stages, sort what you've found into three buckets. This step is where a lot of investigations lose discipline and start guessing instead of categorizing.
| 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 into the first two categories and the total now falls within the accepted reconciliation range, the investigation is effectively closed — you were dealing with a documentation gap, not a material loss. If a genuine unexplained gap remains, that's when it becomes a real deviation.
Step 4: Check equipment hold-up specifically
Hold-up volume is worth its own step because it's the single most under-checked cause of a false reconciliation failure, and it isn't 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 doesn't show up unless someone physically checks.
- Compare against the equipment's known hold-up figure from qualification or previous batches of the same product and batch size — a value wildly different from history is itself a finding worth investigating.
- If hold-up hasn't 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 before you conclude
Numbers only tell part of the story. A five-minute conversation with the operator who ran the stage in question often surfaces something the paperwork never captured — a spill that was cleaned up and reported verbally but never logged, a rework that happened off the record, an extra sample pulled for a customer complaint investigation. Investigations that skip this step and rely purely on records tend to close with "cause could not be determined," which satisfies no one and leaves the same gap likely to recur.
Step 6: Write the root cause, not just the finding
There's 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: why wasn't the withdrawal recorded? Missing SOP step, unclear responsibility, or a form that doesn't have a field for it? The CAPA should fix that gap, not just correct this one batch's paperwork.
A quick pre-investigation checklist
- Verify the theoretical yield and batch size used for comparison are correct
- Confirm all sub-lots and split containers were included in the total
- Work backward: packing → coating → compression → granulation
- Sort findings into accountable loss, process loss, and unexplained loss
- Check equipment hold-up against historical values
- Talk to the operators before writing the conclusion
- State a root cause, not just a description of the gap
If you want to understand the loss mechanisms behind each stage in more depth — not just how to investigate them but why they happen in the first place — see Understanding Yield Loss Points Across Granulation, Compression, Coating, and Packing.