Year-End Stock-Take for Singapore SMEs: Count in October, Not January

If you sell physical goods in Singapore, your year-end stock-take should happen in the last week of October — before 11.11 inbound stock lands and before the festive rush distorts every shelf — not on 31 December, and definitely not in the second week of January. An October count gives you a clean baseline to measure peak-season shrinkage against, leaves enough runway to write off dead and obsolete stock inside the same financial year, and produces a closing-inventory figure your accountant can work with months before it is needed. A January count tells you what you lost. It just tells you too late to do anything about it except record it.

Why is a January stock-take the most expensive one you can run?

Three reasons, and they compound.

The people who can explain the variances have left. Seasonal packers, warehouse temps and part-time retail staff hired for the October-to-December peak are typically off the payroll by mid-January. When a count throws up 40 missing units of a fast-moving SKU, the person who received that shipment, picked it, or wrote it off as damaged is no longer contactable. An unexplained variance is not a finding; it is a write-off.

The audit trail is three months deep. Between an October count and a January count sits 11.11, Black Friday, Christmas and the pre-Chinese New Year restock. That is your highest transaction volume of the year. Tracing a single discrepancy back through that period costs multiples of what the same trace costs in a quiet week.

You lose the ability to act inside the financial year. For a company with a 31 December financial year end, obsolete stock identified in January belongs to the next year's numbers, not the one you are closing. A count in late October still leaves you room to clear slow movers through the 11.11 and year-end sales channels at something above scrap value — which is a commercial decision, not an accounting one. Miss that window and you are carrying the same stock into a Chinese New Year shutdown, paying for the storage, and writing it down anyway twelve months later.

What makes a closing-inventory figure actually defensible?

Under FRS 2 (Inventories), inventory is measured at the lower of cost and net realisable value. In practice, most SME disputes with an auditor or a reviewer come down to four things:

• A consistent cost formula. FIFO or weighted average cost, applied the same way year on year. LIFO is not permitted. If you switched systems mid-year and the new one defaults to a different method, that is a disclosure issue, not a rounding issue.
• Net realisable value, honestly applied. Stock you know you will discount to clear should not sit on the balance sheet at full cost. Selling price less the costs to complete and sell is the test.
• Clean cut-off. This is where most SME counts fail. Goods in transit at the count date belong to whoever holds title under the Incoterms on the purchase order — not whoever has the boxes. Goods received but not yet invoiced still need to be counted and accrued. Consignment stock sitting in a retailer's store is yours; stock sitting in your warehouse on consignment from a supplier is not.
• Evidence that survives. Count sheets with the counter's name and date, a documented second count on every material variance, and a written explanation for anything above your threshold. IRAS requires business records to be kept for five years, and inventory working papers are part of that.

If your closing-inventory number is a single figure exported from a system with no count behind it, you do not have a valuation. You have an assumption.

How do you separate real shrinkage from bad data?

Most SMEs treat every negative variance as theft. In our experience the split is closer to the reverse — the majority of variances in a first proper count are data problems, not losses. The two behave differently, and you can tell them apart before you accuse anyone.

Data errors produce variances in both directions. If SKU A is short 30 units and the near-identical SKU B is over by 28, you have a picking or receiving substitution, not a thief. The same pattern shows up with unit-of-measure errors (cartons booked as pieces), bundles and kits whose components are never decremented, customer returns that are refunded but never restocked, and marketplace integrations that write back sales on a delay.

Real shrinkage is directionally consistent and concentrated. It shows up as steady negative variances on small, high-value, easily pocketed items, and it repeats count after count.

Two pieces of arithmetic are worth doing before anything else. First, express shrinkage as a percentage: (book quantity − counted quantity) × unit cost, divided by cost of goods sold for the period. That single ratio, tracked each year, tells you more than any absolute dollar figure. Second — and this is the one that saves the most time — rank your variances by dollar value, not by unit count. A 500-unit variance on a $0.30 component is a housekeeping matter. A 4-unit variance on a $900 item is the one to investigate. Work the top 20 by value and stop; chasing the tail costs more than it recovers.

Which counting tools are worth paying for under 5,000 SKUs?

Tool selection should follow SKU count and value concentration, not ambition.

Under roughly 500 SKUs: a spreadsheet and a disciplined blind count is genuinely adequate. Blind means the count sheet shows no expected quantity — if the counter can see the system figure, you are testing their eyesight, not your inventory. Two counters, independent sheets, reconcile after.

500 to 5,000 SKUs: a barcode scanning app that writes directly into your existing accounting or inventory system. The scanner is not the point; the integration is. A count captured on a device and then retyped into the ledger has simply moved the error.

Above that, or multi-location: a proper inventory module or lightweight WMS, with cycle counting built in so you never run a full stop-the-warehouse count again.

Whatever tier you land in, five capabilities separate a tool that helps from one that generates work: blind count support, an automatic variance report with a value-ranked view, offline capture (warehouse Wi-Fi fails precisely where the racking is densest), a per-user audit trail of who counted what, and the ability to freeze transactions for the duration of the count. If a vendor cannot demonstrate all five in a live demo, it is a stock list, not a stock-take tool.

One funding note: inventory management solutions have historically appeared on the pre-approved list for the Productivity Solutions Grant. Support levels and the approved-vendor list change, so check the current position on the GoBusiness portal before you build it into a budget — and start early, because approval and deployment lead times rarely fit inside the quarter you decided to act.

What does a workable plan look like from here?

1. Week of 22 September — clean the master data. Deduplicate SKUs, fix units of measure, retire codes with no movement in 18 months. Counting against a dirty master file guarantees variances that mean nothing.
2. Week of 6 October — freeze new SKU creation and pre-count slow movers and bulk storage. Anything that has not moved since June can be counted now and sealed.
3. Week of 20 October — brief the counters. Written instructions, assigned zones, blind sheets, a named person for cut-off decisions.
4. 25–26 October — the full count, with receiving and despatch stopped for the duration.
5. 27–31 October — recount and explain. Second count on every variance above your value threshold; written explanation or a documented write-off for each.
6. November onwards — weekly cycle counts on your top 50 SKUs by value through 11.11 and the year-end peak, with a roll-forward reconciliation at 31 December rather than a second full count.

That last step is what converts an annual ordeal into a control. Once you have one clean baseline and a cycle-count rhythm, you stop shutting the warehouse for a weekend every year — and you find out about a loss in the week it happens, not the quarter after.

Frequently asked questions

Do Singapore SMEs legally have to perform a physical stock-take?
There is no standalone statute that says "count your stock on this date." But the Companies Act requires financial statements that give a true and fair view, FRS 2 requires inventory at the lower of cost and net realisable value, and IRAS expects records supporting your tax computation to be retained for five years. A physical count is the only practical way to substantiate the closing-inventory figure those obligations rest on. If your company is audited, expect the auditor to want to attend the count.

What is a normal shrinkage rate for a Singapore SME?
It varies enormously by sector — high-value electronics and cosmetics behave nothing like industrial spares. Rather than chasing a benchmark, measure your own rate consistently (shrinkage value divided by cost of goods sold), track it across periods, and treat the trend as the signal. A rate that doubles after a peak season staffed by temporary hires is telling you something specific about receiving and despatch controls, regardless of where it sits against an industry average.

Can we count during business hours instead of shutting down for a weekend?
Yes, if you count by zone and freeze movement zone by zone rather than site-wide, and if your system can flag any transaction touching a frozen location. Without that control the count becomes a moving target and every variance is arguable. For most SMEs the pragmatic answer is a one-off weekend count to establish a trustworthy baseline, then a move to cycle counting during normal hours from that point forward.