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

If your financial year ends on 31 December, the right time to run your year-end stock take is October, not the last week of December. Counting in October gives you a clean baseline before the Deepavali-to-Chinese-New-Year order surge, time to investigate variances while the paperwork is still findable, and a closing stock figure your accountant can work with in January instead of March. A December shutdown count does the opposite: it lands in your busiest fortnight, produces numbers nobody has time to verify, and quietly becomes the weakest line in your financial statements.

This matters more in Q4 2026 than usual. Chinese New Year falls in February 2027, which means your purchase orders for the season are placed in November and December — on the strength of stock figures you either trust or you don't.

Why does a December stock take go wrong for Singapore SMEs?

The failure is structural, not a matter of effort. A December count competes with the peak. Your storeman is picking orders, your admin executive is chasing delivery slots, and the count gets compressed into a Sunday. Under that pressure, three things happen.

First, counters record what the system says rather than what the shelf holds, because a count sheet pre-printed with quantities invites confirmation rather than counting. Second, goods in transit — delivered but not yet received into the system, or invoiced but not yet shipped — get double-counted or missed entirely, because nobody freezes the cut-off. Third, the variances that do surface get written off in a single journal entry with no investigation, because the alternative is spending January on it.

The result is a closing stock figure that is defensible on paper and useless in practice. You cannot use it to decide what to order for February.

What does inventory accuracy have to do with your year-end filings?

More than most owners expect. Closing stock sits directly in your cost of goods sold calculation, which drives your profit figure. That profit figure flows into three separate obligations.

• ECI. Estimated Chargeable Income is due within three months of your financial year end — so 31 March for a December year end — unless you qualify for the waiver (annual revenue not exceeding S$5 million and nil ECI). A stock figure you don't believe produces an estimate you don't believe.
• AGM and annual return. Private companies must hold their AGM within six months of financial year end and file the annual return with ACRA within seven months. Both depend on financial statements that depend on the count.
• Record-keeping. GST-registered businesses must retain supporting records for five years. Count sheets, variance approvals and write-off authorisations are part of that trail.

None of this is new law. What is new is how little slack you have: the AGM, annual return, ECI and IR8A work all pile into the same November-to-March window, and the stock take is the one item you can move earlier without asking anyone's permission.

How do you count without closing the warehouse for a day?

Stop treating the count as a single event. Move to cycle counting — counting a slice of your inventory every week, so that everything gets counted at least once over a quarter and your fastest-moving items get counted monthly.

The standard approach is ABC classification. Rank your SKUs by annual value moved. The top 20% by value — your A items — get counted monthly. The middle band gets counted quarterly. The long tail of slow-movers gets counted once a year. A team of two can clear an A-item cycle in ninety minutes before opening, and you never shut down.

Three rules make it work. Count blind: issue sheets with SKU and location but no expected quantity. Freeze movement for the bin being counted, not the whole warehouse. And set a variance threshold — say 2% by value — above which a recount is mandatory before anyone adjusts the system.

Which systems need to agree before you count?

Counting is the easy half. The harder question is what you are counting against, and for most Singapore SMEs the answer is several disagreeing systems.

A typical multi-channel setup has stock moving through a POS at the outlet, a spreadsheet in the warehouse, WhatsApp orders from trade customers, and an accounting system that learns about all of it when someone keys in an invoice. Each hand-off is a re-typing, and each re-typing is a variance waiting to appear in your count.

Before you count, close the loops that create phantom stock:

1. Sales to inventory. A POS sale or a confirmed WhatsApp order should decrement stock at the point of confirmation, not when the invoice is raised days later.
2. Purchases to inventory. Goods received should be booked against the PO on the loading bay, from a phone, with the delivery order photographed and attached — not from a stack of DOs processed on Friday.
3. Inventory to accounting. Stock movements should post to your accounting system automatically. If someone is keying the same figure twice, you have two versions of the truth and no way to tell which one is wrong.

What does the before-and-after actually look like?

Take a wholesale distributor with roughly 900 SKUs across a warehouse and one showroom. Before: a full physical count on the second-last Sunday of December, eight staff, one day of no despatch, variances of 6-9% by value written off in a single line, and a closing figure delivered to the accountant in late February.

After moving to weekly A-item cycle counts from October, with goods receipt captured on the bay and WhatsApp orders flowing into the same stock ledger as the POS: variance drops into the 1-2% range because errors are caught within days of being made, despatch never stops, and the December count becomes a confirmation of about 120 high-value SKUs rather than a shutdown. The accountant gets a closing figure in the first week of January.

What should you do in the next two weeks?

Run your ABC ranking from the last twelve months of sales — most systems will export this, and a spreadsheet pivot will do it if yours won't. Pick your A items. Schedule the first blind cycle count for a weekday morning in the first half of September. Then, separately, list every point in your order-to-cash flow where a number is typed in twice, and fix the highest-volume one before November. The count tells you where you stand; the integration is what stops you drifting again by March.

Frequently asked questions

If we cycle count all year, do we still need a full physical count?

Your auditor or accountant may still want a full count at year end, and if your company is subject to audit you should confirm the requirement with them early rather than assume. But a business with a documented cycle-count programme and consistently low variances is in a far stronger position to negotiate a reduced year-end count, and the count itself takes a fraction of the time because the records are already trustworthy.

We hold stock in three locations. How do we count without stopping sales?

Count location by location, and freeze only the bin or aisle being counted. The critical control is the cut-off: nothing moves in or out of that bin during its count window, and any goods in transit between your own locations are counted at one end only, with the transfer document as the tie-breaker. Do not attempt to count all three sites on the same day — that is the shutdown you are trying to avoid.

What if the count reveals a large shortfall?

Investigate before you adjust. Recount the affected SKUs, then check the obvious causes in order: unrecorded transfers between locations, goods received but not booked in, sales invoiced but not decremented, and unit-of-measure errors where a carton was counted as a piece. Genuine trade losses are normally deductible, but they need documentation — count sheets, the investigation notes and a written write-off approval. Speak to your tax agent before the adjustment is posted, not after.

Digital Perpetual helps Singapore SMEs connect POS, inventory, WhatsApp ordering and accounting so the same number stops being typed three times. If your Q4 purchasing decisions are about to be made on a stock figure you don't trust, that is the conversation to have in September.