Every retailer eventually discovers that the number on the screen and the number on the shelf disagree. A physical inventory count is how you settle the argument. You count what is really there, compare it with what the system says, and correct the system. Done well, it takes a day or a weekend and leaves you with numbers you can trust. Done badly, it makes the data worse than before you started.
This guide walks through a full-store count from preparation to posting the result. It is written for owners and managers who do not have a dedicated inventory team, and it applies to any POS. Where it describes how Retailer OS handles a count, it says so.
Why counts drift in the first place
Before you count, it helps to know what you are looking for. Most differences come from a short list of causes:
- Receiving mistakes. A delivery was short, or was received against the wrong item or the wrong quantity.
- Selling the wrong item. A cashier scans a similar product, so one item goes negative and the other stays high.
- Unrecorded damage and waste. Broken, expired or sampled goods leave the shelf without a record.
- Theft and loss. Customer or employee theft shows up only as a gap between the system and the shelf.
- Unrecorded transfers. Stock moved between stores or to the stockroom without a transfer entry.
- Returns handled loosely. A returned item goes back on the shelf, or into the bin, without the system knowing.
A count does not fix these causes. It resets the numbers so you can see which causes keep reappearing. See how cycle counting reduces shrinkage for the ongoing habit that follows a big count.
Step 1: Choose the date and the scope
Pick a time when the store is quiet. Many retailers count after closing, on a Sunday, or during a planned closure. Decide whether you are counting the whole location or one area. A small store can count everything in a day. A large store is usually easier to handle in sections, one warehouse or stockroom at a time.
Tell the team in advance and assign each person a section. Two people per section is a good rule: one counts, one records. If you are short-staffed, a single person can still count; it just takes longer.
Step 2: Prepare the floor before count day
Preparation decides how accurate the count will be. In the week before:
- Receive everything. Process outstanding deliveries and purchase orders so the system matches the stock that is physically in the building.
- Clear pending transfers. Finish or cancel transfers so no stock is sitting in transit between locations when you count.
- Tidy and consolidate. Put every item in its proper place, group like items together and face labels outward.
- Separate what should not be counted. Mark damaged goods, returns waiting for the vendor, and items on hold, so they are counted on purpose rather than by accident.
- Check barcodes and labels. Replace unreadable labels now rather than on count day.
If you are not sure whether your catalog is clean, fix it first. Duplicate items and inconsistent names are the quiet reason many counts go wrong. See importing products with a clean CSV for catalog hygiene that applies to counts as well.
Step 3: Freeze movement while you count
The count is a snapshot, so stock must not move while you take it. If you sell while counting, the numbers you record will be out of date by the time you finish. The simplest way is to count while the store is closed. If you cannot close, count a section at a time and keep a note of anything sold from that section while you were counting, so you can adjust for it.
In Retailer OS, starting a count takes a snapshot of every stock level in the chosen warehouse. Only one count can be open for a warehouse at a time, which prevents two overlapping counts from applying the same correction twice. Because it is a snapshot, the cleanest approach is to hold sales for the section you are counting.
Step 4: Count by section, not by item
Walk the store in a fixed order, shelf by shelf, aisle by aisle. Counting by location is faster and more accurate than hunting for individual items, because you only touch each shelf once. A few habits help:
- Count the whole shelf, then move on. Do not come back later for a few items.
- Count in the same unit the system uses. If the system tracks single units and you count cases, convert before you record.
- Count open boxes and backstock too. Overflow stock is the most common source of missing units.
- Mark counted areas. A sticker or tag on each finished shelf prevents double counting and missed shelves.
- Recount anything odd. If a quantity looks too high or too low, count it again on the spot before moving on.
Record the quantities as you go. Paper sheets work for small counts. For larger counts, enter the quantities directly into the system rather than copying them later, because copying is where typing errors creep in.
Step 5: Enter the counts and review the variance
Once the quantities are in, the interesting part begins: the variance. Variance is the difference between what the system expected and what you counted. In Retailer OS, a count lists every item with its expected quantity, lets you save the counted quantities in bulk, and shows the variance for each line before anything is changed.
Do not accept every variance blindly. Sort the list and look at the largest ones first, by quantity and by value. For each large difference, ask a few questions:
- Is it a unit mix-up, such as cases versus singles?
- Is the item also stocked in another area that was not counted?
- Was there a recent receipt, return or transfer that was not entered?
- Could a similar item have been scanned instead?
Recount the suspicious lines. A second count settles most of them. What remains are real differences, and those are worth investigating after the count rather than during it.
Step 6: Post the count
Posting is the step that changes your inventory. In Retailer OS, committing a count applies the variance through the normal stock-adjustment process, marks the count as committed and gives you a summary. Because it moves real stock numbers and costs, it is permission-controlled: only users with the right to adjust stock can do it. See roles and permissions for how to limit that to managers.
If you decide the count is not reliable, you can cancel it. Cancelling abandons the count and moves no inventory, so you can start over without any damage. That safety net is worth using. A bad count that is posted is worse than no count at all.
Step 7: Learn from the result
The count is not finished when the numbers are posted. The value is in the pattern. Look at the variance by category and by cause:
- Small, scattered differences usually point to scanning and receiving habits.
- Large losses in one category, especially small, high-value or easily pocketed items, can point to theft.
- Consistent overages suggest receiving or unit-of-measure errors.
- Differences concentrated in one location suggest a process problem at that store.
Write down what you found and what you will change. Then schedule the next count. For many stores, a full count once or twice a year plus regular counts of fast-moving or high-value items keeps the numbers close to reality.
Counting across several locations
With more than one store, stagger the counts rather than doing them all at once, and finish any transfers between stores first. Counting one location at a time lets you borrow experienced counters and compare results. Retailer OS keeps stock per location, so a count in one warehouse does not touch the numbers in another. For the day-to-day side of moving stock between stores, see multi-location inventory management.
Equipment that makes counting faster
You can count with a clipboard, but a few tools speed things up and reduce errors: a barcode scanner for identifying items, a tablet or laptop for direct entry, and printed shelf labels. Which of those you need depends on the size of your catalog. The POS hardware checklist covers what to buy and what you can skip.
A one-page count-day checklist
- Date chosen, team told, sections assigned.
- Deliveries received and transfers finished.
- Shelves tidied, damaged and held stock separated.
- Sales held or tracked during the count.
- Sections counted in a fixed order and marked when done.
- Quantities entered, large variances recounted.
- Count posted by an authorized manager, or cancelled if unreliable.
- Causes written down and the next count scheduled.
Try it on a small area first
If you have never run a count in your POS, start with one stockroom or a single category. You will learn how the screens work, how long counting takes your team and where your labels or catalog need fixing, all before a full-store count. Retailer OS includes a 14-day free trial with no card required, which is plenty of time to practise on a small section.
For the bigger picture, see the complete guide to choosing a retail POS system.
Last updated October 5, 2026