Shopify inventory counts: native tools, count apps, and reconciliation

Counting in Shopify is a solved problem right up to the point where the count is applied. The native count tool scopes a count, lets staff count and scan without touching live stock, then writes the differences as adjustments when you approve. What is not solved is everything after: deciding what a variance means, whether the count was performed honestly, and how the corrected number gets back into the math that drives your buying.

This guide covers the three layers — native counts, third-party counting apps, and reconciliation — in the order they cause trouble, with the cadence arithmetic for a 500-SKU store.

The native count workflow, in four moves

As of November 2026, the admin's count tool works like this:

1. Start a count and pick the location. Counts are created in the desktop admin. Add products by searching, scanning, or importing a list, or start a blank count and let counters add what they find.
2. Count without touching live stock. Staff count in the admin or in Shopify POS. Counting in POS requires a POS Pro subscription. Barcode scanning works with a scanner or the device camera, so a count can be a scan session rather than a paper sheet. Quantities entered during counting do not change the storefront, so the store keeps selling while you count.
3. Review counted versus expected. Shopify presents the comparison per variant before anything changes. This is the review stage, and it is where a good count program spends its judgement.
4. Apply. Approving the count writes adjustments that bring the recorded quantities in line with the counted ones, logged in each variant's history with the date and the person who applied them.

Two operational rules come with that workflow, and both are learned the hard way:

Apply at close of day. Applying a count mid-afternoon overwrites the units that sold while you counted. A store that counts 41 mugs at 11am and applies immediately erases the six that sold between open and 11am — which then reappear as either fake shrinkage or phantom stock the storefront keeps selling. Count after the last sale; apply before trading resumes.

Freeze receiving for the count window. A purchase order received halfway through a count injects stock into a session that is supposed to measure what was already there. Receive first, then count, or hold the delivery.

Documented limits

These are product limits, not bugs, and they should shape the routine:

• Counts are created on desktop. You cannot create a stock count in the Shopify mobile app or in Shopify POS. The counting itself can happen in POS, but the session starts in the admin.
• POS counting needs POS Pro. If your team counts on a POS device, confirm the subscription before you plan a full stock take around it.
• Counts cannot be exported. Shopify does not provide a CSV of count results. Copy the variances onto your own sheet before you approve, or you will lose the trail.
• One location per count. A count is scoped to a location. A multi-location store runs separate counts, or counts one location at a time.
• Bundles are outside the count. Bundle and kit listings are not counted as components in the native count; count the underlying inventory items.
• Duplicate barcodes error out. If two variants share a barcode, scanning is ambiguous and the count will stop or mis-assign. Clean the barcodes before the session.
• Per-product history is 180 days. Older movement needs the reports layer, which is plan-dependent (what each native report does and doesn't cover).
• No variance workflow. Shopify tells you counted versus expected. It does not ask for a cause, assign an owner, or track whether the variance was investigated.
• No recurring schedules. A count carries a planned date, not a repeat interval; the cadence is a calendar entry in your own process, not a Shopify setting.

None of those is a reason to avoid native counting. They are the reasons a native count needs a companion process — and, for some stores, a companion app.

Native, apps, or paper: the honest comparison

On the app tier, check the listing's review count and read the recent reviews rather than trusting a feature list. The category's failure mode is a count that looks complete and isn't: a free-tier variant cap, a silent sync failure, or a scan that assigns to the wrong variant. Verify the variant cap before planning a full stock take around a free plan, and test a small count before you trust a large one.

The reconciliation step (the part that gets skipped)

Counting produces numbers. Reconciliation produces understanding. Three moves, in order, and the middle one is the one that gets dropped:

1. Compare. Counted versus expected per variant, with the dollar value of each variance at cost — not just the unit count. A −3 on a $4 item and a −3 on a $90 item are not the same finding, and a list sorted by units will always put the wrong one on top.

2. Explain. Before adjusting, check the variant's recent trail: any receipt in the last two weeks? A transfer? A rush of till sales at one location? A bulk edit? Most large variances trace to a single event, and the answer is usually receiving. A worked example: a shelf shows −5 against the book; the PO for that SKU was received four days ago as 24 units when the carton held 19. The variance isn't shrinkage — the book was born wrong, and the fix is to correct the receipt, not to adjust the stock.

3. Adjust and close. Apply the correction so the system matches the shelf, with the reason Count for a count finding, or the appropriate reason for a known cause (Damaged for unsellable units, Correction for a receipt keyed wrong). Then close the loop in writing: SKU, date, counted versus expected, cause, fix. The full routine — including how to reconcile Shopify against a POS, a 3PL, and a spreadsheet — belongs to the stock reconciliation guide; the count is only its input.

A variance closed without a cause is not closed. It is scheduled to recur, and next quarter it will arrive with the same amount of surprise.

A cadence with arithmetic attached

The cadence question has an answer, and it is arithmetic rather than philosophy. Grade your catalog by selling rate — Shopify's native ABC analysis grades variants A, B, and C by sales, and it is the natural input. Then count A monthly, B quarterly, C twice a year:

Ninety minutes a week, usually split across two quiet mornings, and the annual stock take stops being a crisis. Re-grade quarterly: a new bestseller stuck on the twice-yearly list is the most expensive scheduling mistake in the system, and a faded A item still on the monthly list wastes the hour you needed for it.

Measuring whether the counting is working

Every count produces an accuracy rate, and the rate is more useful than any single variance. Compute it line-level — of the SKUs counted, how many came back exactly right — and track it monthly. A store at 99% has a tight process and noise-level misses; a store at 93% has a structural cause, almost always receiving, and should stop trusting its reorder math until it is fixed. The practical reading is that your A items should count clean nearly every time, and the trend should be upward.

What counting cannot tell you

A count certifies the number on count day. It says nothing about the six days after, which is where the expensive failures live: a bestseller selling out on Tuesday with the next count on Friday, a variant that sat dark for a week recording zero sales that look exactly like zero demand. Counts measure accuracy; they cannot measure awareness. That gap is a monitoring problem, and it is the reason a store can be 99% accurate and still lose its best week of the quarter.

Common questions

Does Shopify have native inventory counts? Yes. The admin includes a count tool. You create the count on desktop, add products by search, scan, or list, and count in the admin or in Shopify POS with a POS Pro subscription. Quantities only change when the count is reviewed and applied. Counts cannot be exported, so record the variances on your own sheet before approving.

How often should I count inventory in Shopify? Grade by sales velocity and count A items monthly, B quarterly, and C twice a year, with a full stock take annually to reset the baseline. On a 500-SKU store that is roughly 45 SKUs a week, or about 90 minutes.

What's the difference between a count and a reconciliation? A count measures what is physically there. Reconciliation compares that to the book, explains any variance to a documented cause, and adjusts with a reason attached. Skipping reconciliation means fixing today's number while leaving the cause active.

Should I count before opening or after closing? After the last sale of the day, and apply before trading resumes. Applying a count taken during trading hours overwrites the units sold while you counted, which shows up as phantom shrinkage or phantom stock.

Can I count with a barcode scanner? Yes — scanning works in the admin and in Shopify POS, with a scanner or the device camera. Dedicated scanners and count apps are faster for long sessions and multi-person counts, at roughly $15–30/month.

Can I export a count? Not natively. Shopify does not provide a CSV of count results at the time of writing, so copy the counted-versus-expected variances onto your own sheet before you approve the count.

Next step

Open the count tool today and run one session on your fastest-moving shelf, at close of day, and apply it before you leave. Then write the variance down on one line: SKU, counted versus expected, and the cause. That single line is the habit the rest of the system is built on — and if you want the annual version planned properly, the stock take routine is the next read.