Doing a pharmacy stock count that actually reconciles

Why counts do not reconcile

Three reasons, in order of frequency:

1. Stock moved during the count. Somebody dispensed from a shelf that had already been counted, or one that had not.
2. The count was by medicine, not by batch. Forty tablets on the shelf across two batches, entered as forty against one, and the expiry picture is now wrong even though the total is right.
3. Somebody entered what they expected. The system said 40, the shelf looked about right, 40 was entered. This is the one that hides real losses for years.

The method

Before

• Pick a time when the pharmacy is closed. Early morning beats evening — people are more accurate before a shift than after one.
• Print the expected quantities by batch. Or do not, and do a blind count — see below.
• Freeze dispensing. Nothing leaves the shelf until the count is finished.
• Put expired stock in a separate box first. It should not be on the shelf anyway.

During

• Count by batch. Batch number and expiry date, then quantity. This is the whole point — a total without batches tells you nothing about what is about to expire.
• Two people for high-value items. One counts, one records. Not for everything; for the top twenty by value.
• Count opened packs honestly. Seven tablets left in a strip of ten is seven, not one strip.
• Do not stop to investigate. Note the discrepancy and keep counting. Investigating mid-count is how counts take four hours.

After

• Enter what you found, not what you expected.
• Record every difference as an adjustment with a reason — damaged, expired, unrecorded dispense, counting error, unknown. "Unknown" is a valid and useful reason.
• Investigate anything above your threshold, not everything.
• Unfreeze dispensing.

Blind counts

Counting without the expected figure in front of you is more accurate and slightly slower. If your discrepancies are suspiciously small and suspiciously rare, you are not doing a count — you are confirming a number. Do one blind count a year at minimum.

Why adjustments must be entries, not edits

If a count "corrects" the quantity by overwriting it, the discrepancy disappears along with the evidence. Recorded as an adjustment in anappend-only ledger, it accumulates into the most useful report a pharmacy has: adjustments by reason, by month, by medicine.

A steady drip of small negative adjustments on a few high-value items is a pattern. It is invisible if you edit numbers and obvious if you record movements. This is one of theseven places clinic revenue leaks.

How often

• Full count — twice a year, and at financial year end.
• Cycle counts — a small set weekly. Twenty items, rotating, takes fifteen minutes and finds problems months before the full count does.
• High-value items — monthly, regardless of the cycle.

What a reasonable discrepancy looks like

Under 1% of stock value is normal. Over 3% needs an explanation. Consistent losses on the same few items are not shrinkage in general — they are shrinkage in particular, and worth a quiet look at who dispenses them.

Clinikr derives stock from an append-only ledger, so a count is an adjustment entry with a reason attached, and adjustments can be reported by reason and by medicine.