After closing time, a clerk scans the last shelf on a phone and the screen shows: the system says 24, the shelf holds 21. Where did the three go: lost, misrung, or never entered at receiving? For most convenience stores, the hard part of how to count inventory is not the counting itself. It is what to do once the numbers do not match.
This piece is for stores that have crossed 300 SKUs and now lose selling hours to counts. Everything below is something you can change tonight.
Who this is for: past 300 SKUs, counts eating into open hours
Shopify's help center publishes staffing estimates by catalog size: under 500 variants takes one person two to four hours; 500 to 2,000 variants takes two people most of a day; 2,000 to 5,000 takes two or three people a full day; above 5,000 means zoning across several days.
The implication is direct. As SKUs climb from 300 to 800, counting moves from a quick task after close to a scheduled shift. When owners say counting got harder, it is rarely laziness. The store outgrew the method that worked at 300 SKUs. If you run one store alone, read the process section. If you already run three shifts with around ten staff on one system, read the cadence and permissions part.
What does inventory counting mean and what methods are there?
Start with three numbers. Book stock is what the system records. Physical stock is what sits on the shelf and in the back room. A count variance is the gap between them. Counting is not about the goods; it is about the gap between those three numbers. The first number is the one shops get wrong most often, because many small stores keep book stock as a figure typed in after the last count, not as a record built from documents.
There are three common methods. A full count, also called a shutdown count, counts the whole store at once. It is accurate but needs the doors closed. The cycle counting method splits the work by zone or ABC class, counting part of the catalog each day or week until a full year covers everything. ABC ranking puts high-value and fast-moving items on a weekly cadence and slow, low-value items on a quarterly one.
One figure gets ignored too often. The National Retail Federation (NRF) and the Loss Prevention Research Council's 2023 National Retail Security Survey covered fiscal 2022 across 177 retail brands. Retailers attributed shrinkage to external theft at 36%, internal theft at 29%, process and control failures at 27%, and unknown at 6%.
That means roughly a quarter of all variance is not stolen. It is short shipments never caught, returns booked to the wrong SKU, transfers between stores never invoiced, damage never logged. Counting catches exactly this slice, and fixing it costs the least. Assume the record is wrong first; investigate theft only after the records check out. The return on effort is different.
How to choose between cycle counting and a full shutdown count?
The choice depends on SKU count and shift count, not on which method is newer.
A shutdown count counts the whole store at once and is the most accurate, at the cost of sales during the closure. Shopify puts it plainly: with no sales happening during the count, a closed-store count is the most accurate option, but the store loses revenue while closed.
The cycle counting method weaves the count into the daily shift. Shopify's frequency guidance: quarterly works for most retailers; monthly suits high-shrink small high-value goods; a rotating weekly zone count covers the whole catalog without a store-wide shutdown.
The rule of thumb: below 500 SKUs with one shift, a shutdown count twice a year is simplest. Above 500 SKUs with two or three shifts, move to cycle counting, with weekly counts for high-value and fast movers and quarterly for the slow ones.
Overseas Chinese wholesalers hit a different constraint. On Bangkok's Sampeng Lane and in Jakarta's Glodok, a stall with a 200+ SKU barcode catalog may have the boss keeping books in Chinese while staff receive goods in Thai or Indonesian, with 500-plus slips a day. A full shutdown is nearly impossible, and the numbers drift again the next morning. The same pattern shows up across Southeast Asia and the Gulf.
Four mistakes new stores make
First, counting without freezing transactions. Selling while counting means the numbers never reconcile. Book all pending orders, transfers, and receipts before you start.
Second, editing the number without finding the cause. The variance is a symptom. Changing book stock to match physical stock erases the symptom, and the same error returns next month.
Third, sending everyone in at once with no clear ownership. Zone by physical location, one person per zone, sign off when done.
Fourth, counting the front and skipping the back room. The warehouse count follows different rules. The floor shows shelf display; the warehouse shows receiving documents, damage logs, and the returns area.
A count process that actually runs: zone, freeze, first count, recount, assign
Step one, prepare the system. Clear pending orders, transfers, and receipts so stock reflects the moment before the count. Step two, draw zones by shelf, cooler, and warehouse, small enough that one person finishes one zone in a single pass. Step three, freeze transfers. Inside the window, pause cross-store and cross-warehouse moves, or the same goods get counted twice. Step four, first count by scan, not by name search; pre-print labels for items with no barcode. Step five, recount. Re-count any variant whose variance exceeds the threshold, commonly 5%. Step six, assign a reason code to every variance: short receipt, entry error, wrong-SKU return, unreported damage, unrecorded transfer, other. Watch the monthly mix and fix the process that is growing.
Many owners first look for an inventory count sheet template, but a sheet records how much was off, not who changed what, when, and why. The line between an inventory system and an Excel count is traceability.
One more lesson for multi-store operators: when one store shows a positive variance and another a negative one, the usual cause is a transfer between them that was never recorded.
Where do you start when inventory discrepancies can't be traced?
Check the receiving document first, not the camera footage. Fix the order as five steps and do not skip.
1. Receiving documents. A short or extra shipment from the supplier is the most common starting point, especially for bulk-weight and mixed-SKU goods.
2. Sales returns. A return booked to the wrong SKU throws two locations off at once.
3. Cross-store transfers. Goods left, document not opened, and the books go one positive, one negative.
4. Damage and write-offs. Near-expiry, breakage, samples, anything unlogged becomes variance.
5. The barcode itself. Two SKUs sharing one barcode never scan right; the fault is in the item master, not the count.
Only variances these five steps cannot explain move into loss-prevention and internal-theft review. Skipping to suspecting staff before step five hurts the team and wastes the most time. How to fix inventory discrepancies is not about flattening the number; it is about leaving a traceable reason code on every line.
What to look for in a counting tool, and a realistic time and cost expectation
Judge a tool on four fit conditions, not a feature list. First, can it count by scan on a phone and store the count offline, syncing later? This decides whether counting can happen during business hours. Second, does it carry discrepancy reason codes and adjustment approval: who changed what, when, and why. If those three facts do not persist, you will still be hunting causes next month. Third, is the cross-store transfer a standalone document you can trace on its own? Across multiple stores this is the largest source of variance. Fourth, can staff permissions be separated, with the owner seeing the whole picture and employees seeing only their scoped data? Overseas stores add one more: when the owner and local staff speak different languages, the interface language should follow the person, and a multilingual interface keeps both reading the same screen.
On these criteria, Ailit inventory software maps to conditions two through four. Ailit is an AI-powered intelligent inventory software for SMEs, built by Kingdee — a Hong Kong main board-listed, world-leading SaaS company. Ailit supports Simplified Chinese, Traditional Chinese, English, Spanish, Portuguese, Arabic, Thai, and more languages, serving merchants in more than 130 countries and regions, supporting settlement in 154 currencies with 170+ exchange rate settings. Kingdee's Zhihuiji line serves over 3 million merchants, and Ailit is its international edition.
Set a realistic timeline. Moving stock from sheets into a system and running a full count usually takes one to two weeks to stabilize, and you will find a batch of item-master errors along the way. Once stable, cycle counting compresses to 15 to 30 minutes a day, done by one person on the opening or closing shift.
Shopify's help center earns a line on the count sheet: regular, consistent inventory counts are worth more than infrequent perfect ones.
Frequently Asked Questions
How often should I count inventory?
Usually a full count each quarter, paired with a rotating weekly zone count. Push high-shrink small high-value goods, batteries and beauty samples, to monthly. A store under 500 SKUs on a single shift can accept a count every six months, but log a reason code every time. Knowing how to count inventory is one part; cadence is the other.
What does it take to close a discrepancy?
Usually three steps: re-count any variant over 5% variance, assign a reason code to every line, then review the monthly mix. Changing book stock to match physical stock is not a close, because the same error returns next month. Ailit binds the adjustment to its reason code and keeps the operator and timestamp on the record.
Can you count without closing the store?
Yes, if you move to cycle counting and freeze cross-store transfers during the count window. Split zones across shifts, one person per zone, submit when done. Above 2,000 SKUs with three or more shifts, a zoned count without closing is the only realistic option.
How long can Excel or paper forms hold up?
As long as you need to answer who caused a variance and when, which is exactly where a sheet stops. It cannot record the operator or the reason code, and it has no permission control. Moving from sheets to an inventory system usually takes one to two weeks to run the first full count.
