Skip to content
Online Business

Stock Take for Omnichannel Sellers

SME Academy ·Updated 30 Jul 2026 ·6 min read
Stock Take for Omnichannel Sellers
Key takeaways

Counting stock is easy. Counting stock that is simultaneously listed on three marketplaces, a webstore and a shop counter is a different exercise — because the number you are checking against is being changed by strangers while you count. Cycle counting solves it; annual counting does not.

Counting stock in a single shop is a straightforward job. Counting stock that is simultaneously listed on Shopee, Lazada, TikTok Shop, your own webstore and a shop counter is a genuinely different exercise, and the reason is uncomfortable: the number you are checking against is being changed by strangers while you count. Every sale during the count moves the target. That is why the annual full count that works for a single shop produces meaningless variances for an omnichannel seller — and why the answer is not counting harder, but counting more often on a smaller scale.

Why annual counting fails here

  • The count takes hours and sales do not stop. Unless you close every channel, the system figure moves under you.
  • A year of drift is uninvestigable. A variance of 14 units discovered in December could have happened in March. Nobody will ever know why.
  • It finds the problem far too late. Stock accuracy failures cause overselling and cancellations continuously — see how overselling happens — so a discovery in December is eleven months of penalties you already took.

Cycle counting: count a little, constantly

Instead of counting everything once, count a slice every week so that everything is counted over a cycle. It takes less total time, it is far more accurate, and — the real benefit — a variance found this week has a cause you can still remember.

A workable ABC schedule:

Class What it is Count frequency
A Fast movers and high value — usually a small share of SKUs and most of the revenue Weekly
B Steady sellers Monthly
C Slow movers and low value — the long tail Quarterly

Add three rules on top:

  • Count any SKU immediately after an oversell incident, regardless of class.
  • Count anything with a variance last time, until two clean counts in a row.
  • Count anything that has just come off a promotion. Campaign periods are when records break.

Running a count that produces a usable number

  1. Pick a freeze window. The quietest hour you have — early morning, or after close. You are not trying to stop all sales; you are trying to count during the smallest possible number of them.
  2. Print or export the system figure first, timestamped. Counting first and comparing to a system figure pulled afterwards is the most common way to produce a fake variance.
  3. Count the physical stock, blind. Do not show the counter the expected number. A counter who can see "12" will find twelve.
  4. Count everything, including the places stock hides — the back room, the display shelf, goods received but not put away, and stock allocated to unshipped orders. That last one is where most phantom variances live.
  5. Record on a fixed sheet, one line per SKU, with a column for the physical count and a column for the difference. Our free stock take sheet is set up for this.
  6. Investigate before adjusting. Adjusting the system to match the count hides the cause, and the cause will produce the same variance again next month.

Investigating a variance properly

Work through this list in order — the first four explain most of them and none is theft:

  1. Unshipped orders. Stock sold but still on the shelf, or allocated and physically moved. Check whether your system counts allocated stock as on-hand.
  2. Goods received not recorded, or recorded twice.
  3. Returns not put back into stock, or put back without being recorded.
  4. Bundles and kits that did not decrement their components.
  5. Damage and write-offs that happened but were never recorded.
  6. Miscounts — recount before concluding anything else.
  7. Only then, consider shrinkage.

Whatever the cause, write it down against the SKU. Three months of recorded causes turns stock control from guesswork into a list of specific fixes.

The two numbers that measure whether it is working

  • Stock accuracy = SKUs with zero variance ÷ SKUs counted. Track it monthly; the trend matters more than the level.
  • Variance value = the ringgit value of the differences, positive and negative separately. Netting them off hides the problem — RM500 over and RM500 short is not a clean month, it is two problems that cancelled.

Preventing the drift in the first place

Stock takes measure a problem. These reduce it:

  • One source of truth for stock, with every channel reading from it — see syncing inventory across channels and connecting your POS.
  • Receive properly. Count goods in against the delivery order, at the point of receipt, before they touch a shelf.
  • Record returns the moment they arrive, not when someone gets to them. See returns and refunds.
  • One person owns adjustments. Anyone able to change a stock number without a reason code is a source of drift.
  • Reconcile after every campaign, while the causes are still fresh.

Frequently asked questions

How often should I do a full count?
If cycle counting is running properly, a full count becomes largely a formality — though your accountant may want one at financial year end for the accounts. Ask them what they need rather than assuming a full count is mandatory, and remember that stock valuation for the accounts and stock accuracy for operations are two different purposes.

Do I have to close my online channels to count?
Usually not. Count in the quietest window, snapshot the system figure at the start, and account for anything sold during the count. Closing channels costs sales and is rarely necessary for a cycle count of a few dozen SKUs.

What is an acceptable variance?
The honest answer is that the trend matters more than any threshold, and a benchmark from another business tells you little. Measure your own accuracy, set a target above it, and treat any single large variance as an investigation regardless of the overall percentage.

Should I count stock held at a fulfilment centre or marketplace warehouse?
Yes — it is your stock and it is where discrepancies are hardest to see. Reconcile their reported figure against your records on the same cycle you use for your own shelves, and raise differences promptly; these get harder to resolve with age.

Can I use my phone instead of paper?
Yes, and a phone with barcode scanning is faster and less error-prone than writing. The principle that matters is not the medium — it is counting blind against a timestamped system figure. A phone that shows the expected number defeats the purpose exactly as a printed sheet with the numbers filled in would.


Sources: this guide is inventory-control practice rather than a statement of accounting standards. Requirements for stock counts and stock valuation at financial year end are matters for your accountant and depend on your reporting obligations — confirm what is required for your accounts separately from what is useful for your operations.

Recommended tool
Not sure how much SST you'll owe?

Estimate it in under a minute with our free calculator.

Open calculator
Share:

Related guides