Year-End Stocktake Timing: Count in December, Not 11.11 Week

If your financial year ends on 31 December, do not run your full year-end stocktake now. Cycle count through November, then schedule the full count for the quiet window after the 12.12 dispatch tail clears, which for most Singapore SMEs means roughly 15 to 24 December, before the Christmas and New Year leave empties the warehouse of the people who know where things are. A full count in the seven days before 11.11 is the single least defensible date on the calendar: it freezes picking at the exact hour your orders peak, and every unit you count will have moved several times before 31 December anyway.

Why does a full count before 11.11 cost more than it saves?

A full physical count only produces a trustworthy number if stock stops moving while you count. For a typical SME holding 1,000 to 1,500 active SKUs across one warehouse, two counters need about a day and a half, plus half a day for recounts and variance investigation. That is two days of suppressed picking.

For a Singapore seller trading on Shopee and Lazada, 11.11 week commonly carries 8 to 15 per cent of the whole month's revenue. Losing dispatch capacity in that window does not just defer the sales. Late dispatch pushes your late shipment rate above the marketplace threshold, which costs you listing visibility for the rest of the quarter and can put a Preferred or Mall badge at risk. You pay twice, and you still have to count again in December because the November number will be meaningless by year end.

The only reason to count a full warehouse in early November is a suspected theft or a system failure so severe that you cannot trade on your current figures. That is an incident, not a stocktake.

What should you do in November instead?

Start daily cycle counting this week. Cycle counting means counting a small, rotating slice of locations every day without stopping operations, so that by the time you reach the full count in December, your records are already close and the variances are small enough to explain.

The practical version takes about 20 minutes a day:

• Split your SKUs by value using an ABC rule. The A group is usually the top 20 per cent of SKUs carrying around 80 per cent of stock value. Count A items monthly, B items quarterly, C items once a year.
• Count 15 to 25 bin locations each morning before picking starts, not at the end of the day when the picked-but-not-dispatched pile distorts everything.
• Count by location, not by SKU list. Counting a location tells you what is actually there, including the units sitting somewhere they should not be.
• Record the variance, correct the system the same day, and write one line on why. A variance with no cause recorded will reappear next month.
• Investigate anything over 2 per cent of line value or SGD 200, whichever is lower. Below that, correct and move on.

Through 11.11 itself, suspend counting for the three or four days of heaviest dispatch and resume on the Friday. Counting during a picking surge produces noise, not data.

How do you count stock that sits in a marketplace warehouse?

This is where most SME counts quietly go wrong. If you use Fulfilment by Lazada, Shopee's fulfilment service, a third-party logistics provider or consignment space in a retailer, that stock is still yours and still belongs on your balance sheet, but you cannot physically count it.

For each external location, request a stock-on-hand report dated at your cut-off, then reconcile three figures: what the provider says it holds, what your own system says is there, and what is in transit between you and them. Units in transit on 31 December belong to whoever holds title under the delivery terms, so inbound shipments you have paid for but not received may still be yours to count. Also separate reserved units, meaning stock allocated to orders already placed but not yet shipped, because marketplace dashboards often show these as unavailable rather than as stock you own.

Ask for those reports before mid-December. Provider support queues in the second half of December are slow, and a missing warehouse report is the usual reason a stocktake slips into February.

What does your accountant or auditor actually need?

Most Singapore SMEs are exempt from audit under the small company criteria in the Companies Act, which look at revenue, total assets and headcount. Exempt does not mean the count is optional. Closing stock feeds directly into your cost of goods sold, so the number drives your reported profit and the tax computation behind your Form C-S. An unsupported closing stock figure is a weak position if IRAS ever asks how it was derived.

Keep three things from the count: the dated count sheets with the counters' names, the variance report with explanations, and the reconciliation from count date back to 31 December if you counted early. If you do require an audit, agree the count date with your auditor now, because they may want to attend and their December diary fills fast.

Decide your obsolete and damaged stock write-off before 31 December as well. Stock you know is unsellable should be written down in the year it became unsellable. Carrying it into 2027 overstates this year's profit and the tax that follows it.

What should be settled by Friday 6 November?

Four decisions, and none of them take longer than an hour:

1. Fix the full count date and put it in writing to the warehouse team, your accountant and any auditor. Mid-December, after the 12.12 tail, before leave starts.
2. Publish the cycle count roster for November and December, naming who counts and which zone on which day. An unnamed roster does not happen.
3. Request external stock reports from every marketplace warehouse and logistics provider, with the cut-off date stated.
4. Set the cut-off rules for the full count day: last dispatch time, where received-but-not-put-away goods will sit, and that no stock moves between the count and the sign-off.

Done this way, the December count becomes a verification exercise rather than a discovery exercise. Teams that cycle count through peak typically arrive at year end with inventory accuracy in the high nineties, which turns a three-day shutdown into a one-day confirmation and removes the February scramble to explain a variance nobody can now remember.

Frequently asked questions

Can we skip the full year-end count entirely if we cycle count all year?
If every SKU has been counted within the financial year, variances are documented and your accuracy is consistently above about 97 per cent, many accountants will accept cycle count records in place of a single full count. Agree this with them in advance, in writing. Do not assume it.

We counted in October. Do we need to count again?
Not necessarily, but you must be able to roll the October figure forward to 31 December using complete purchase and sales records, with no unexplained gaps. After a peak season of high volume and high returns, that roll-forward is often harder than simply counting again in December.

How should we treat customer returns that are still unprocessed at year end?
Returns you have accepted but not yet inspected are your stock and must be counted, valued at what you can realistically sell them for rather than full cost. Clear the returns backlog before the count date if you can, because an uninspected returns pile is the most common source of a stock value that cannot be defended.

If you want the cycle count roster, the ABC split and the cut-off rules drawn up against your own SKU list and warehouse layout, that is a half-day piece of work and it pays for itself in the December you do not lose.