How to Do a Stock Take: Step-by-Step Guide

What a stock take is (and why it matters)

A stock take — also called a physical inventory count or stocktaking — is simply counting everything you actually have on hand and comparing it to what your records say you should have. The gap between the two is called shrinkage, and it's caused by things like theft, breakage, supplier short-deliveries, miskeys at the till, and returns that never made it back onto the shelf.

Getting this right isn't busywork. Your stock figure feeds straight into your accounts: it affects your cost of goods sold, your profit, and the value of inventory on your balance sheet. If the number is wrong, your profit is wrong. Accurate counts also stop you from overselling online, running out of bestsellers, or tying up cash in dead stock.

How stock is valued and reported for tax can vary by country. This is general information — confirm the current rules with your country's tax authority or a qualified local tax professional before finalising figures for a return.

Before you start: plan the count

Most failed stock takes fail before anyone counts a single item. A little preparation makes the whole thing faster and far more accurate.

• Pick a quiet time — after close, on a slow day, or at month-end — so stock isn't moving while you count.
• Tidy and label shelves, group like items together, and clear out rubbish and empty boxes.
• Decide your counting unit for each product (each, box, pack, kg) and stick to it.
• Set aside goods that shouldn't be counted: customer layaways, damaged returns, and consignment stock.
• Assign counters and, ideally, pair them up — one counts, one records.
• Prepare your count sheet or open your inventory app on a phone or tablet.

Step-by-step: how to run the stock take

1. 1

   Freeze stock movements

   Stop receiving deliveries and, if possible, pause sales during the count. If you must trade, keep new arrivals in a clearly marked holding area and count them separately at the end so nothing is double-counted or missed.

2. 2

   Print or open your count sheet

   List every item with its location, description, SKU or barcode, and expected quantity. In Tallium you can pull a live stock report per location so your counters have an up-to-date starting list rather than a stale spreadsheet.

3. 3

   Count location by location in a fixed order

   Work systematically — top to bottom, left to right — through one shelf, aisle or storeroom before moving to the next. A consistent path stops you from skipping or recounting sections.

4. 4

   Record every quantity, including zeros

   Write down what you physically see, not what you expect. If a bin is empty, record zero. Scanning barcodes with a phone is faster and cuts transcription errors compared with hand-writing SKUs.

5. 5

   Count goods in transit and holding areas last

   Once the main floor and storeroom are done, count anything set aside — deliveries received during the count, items awaiting return to shelf, and back-office stock — so the total is complete.

6. 6

   Do a blind recount of high-value or high-variance items

   For expensive or fast-moving lines, have a second person recount without seeing the first figure. Where the two disagree, count a third time and use the agreed number.

Reconcile the count against your records

Now compare your counted quantities with your system's expected quantities. The difference — the variance — is where the real work is. Don't blindly overwrite your records; understand the gaps first, because a big variance often points to a process problem, not just missing stock.

In Tallium the counted figures line up against expected on-hand values automatically, so variances are highlighted for you rather than worked out by hand across two spreadsheets.

• Sort variances by value, not just by quantity, so you tackle the costly gaps first.
• For each large gap, check for obvious causes: unentered deliveries, wrong unit of measure, or a miscount.
• Confirm the item wasn't counted under a second SKU or stored in another location.
• Look for offsetting errors — one product up by 10, a similar one down by 10, often means a mix-up.

Post adjustments and update your accounts

Once you're confident the counts are right, post stock adjustments so your system reflects reality. Keep a short note against each significant adjustment explaining the reason — it makes future audits and pattern-spotting much easier.

Because inventory value flows into your accounts, a stock adjustment changes your cost of goods sold and profit. In an all-in-one system like Tallium — where POS, inventory and accounting share one set of records — posting the adjustment updates your stock levels and your books together, so you don't rekey the same numbers twice.

How and when inventory write-downs are recognised can differ by country and can affect your tax position. This is general information — confirm the current treatment with your country's tax authority or a qualified local tax professional.

Move to cycle counting so you rarely shut down again

A full stock take is disruptive. Cycle counting spreads the work out: instead of counting everything once or twice a year, you count a small slice of your stock often — say a handful of items every week — on a rolling schedule.

Prioritise your fastest-moving and highest-value lines for more frequent counts, and count slow, low-value items less often. Over time this keeps your records accurate all year round, catches shrinkage early, and means the annual count becomes a quick confirmation rather than a painful all-nighter.

• Count your top sellers and most valuable items weekly or monthly.
• Rotate through the rest on a schedule so everything is counted a few times a year.
• Track your variance rate over time — a falling rate means your processes are improving.

Common mistakes to avoid

• Counting while stock is still being sold or received — freeze movement first.
• Recording what you expect instead of what's physically there.
• Using different units (a box vs. individual items) between count and system.
• Overwriting records without investigating large variances.
• Leaving it to one tired person late at night — pair up and take breaks.
• Never repeating the process, so errors quietly build up again.