Shopify stock take: a full-count routine a small store can actually finish
A Shopify stock take is a full physical count of every SKU at every location, valued at cost. Here is the routine, a count sheet, and the reconciliation step.
InventoryX is in development and not available today. ClearStock is the live Shopify app.
A stock take is one full physical count of every SKU at every location, taken on a fixed date, valued at cost, and posted back into Shopify before trading resumes. It is not a better version of a cycle count — it is a different instrument. A cycle count keeps a rolling slice of the catalog honest; a stock take resets the whole baseline at once, which is why it belongs on the calendar once a year rather than every month, and why the stores that run it well spend more time preparing than counting.
This is the routine for a small Shopify store: what the day is for, how it differs from cycle counting, the two-week run-up, the count sheet that holds up on the floor, and the reconciliation that turns a pile of counted numbers into an adjustment you can defend.
Stock take vs cycle count: one table, no ambiguity
| Stock take (full count) | Cycle count | |
|---|---|---|
| Scope | Every SKU, every location, including 3PL stock | A scheduled slice — one aisle, one ABC batch |
| Frequency | Yearly, or after a migration/relocation | Weekly to monthly, forever |
| Trading | Usually closed, or counted after close and posted the same night | Business as usual |
| Time at ~500 SKUs | 1–2 staff-days including reconciliation | 60–90 minutes a week |
| Output | A valuation you can hand an accountant, plus a reset baseline | A trend line on accuracy for the SKUs that matter |
| Fails when | It is treated as a data-entry event instead of a reconciliation event | It stops happening in busy months |
The relationship is complementary, not competitive: cycle counts shrink the size of the annual surprise, and the stock take catches what the rotation missed. If your full count keeps producing large, unexplained variances, the problem is the twelve months before it, not the Saturday.
What a stock take is actually for
Four jobs, and only the first is about quantity:
- Reset the baseline. Every uninvestigated discrepancy from the year gets resolved at once, so the next twelve months of reorder math start from a number that was verified physically rather than inherited.
- Produce a valuation. Year-end inventory at cost feeds your balance sheet and cost-of-goods reconciliation. The count, valued at cost, is the support for that number — which is why the count sheet and the valuation sheet should be the same file.
- Support a claim or a lender. Insurance schedules, financing packages, and (in some states) inventory tax filings all rest on a dated record of what you owned, where, and what it cost. A count is the moment that record gets refreshed.
- Test the process, not just the stock. The variance list is a diagnosis of your receiving, your transfers, and your till. A count that finds nothing is either a very good year or a count that was performed carelessly.
One honest note on the third job: a stock take supports a claim, it doesn't replace the paperwork behind it. The dated quantities and the receipt references are what an adjuster asks for, which is why the count sheet carries a receipt reference column at all.
The run-up: two weeks, four decisions
The count itself is the easy part. These are the decisions that determine whether the number you post means anything.
T-14 — pick a date that is not inside a receiving window. The single most common way a full count produces fake variances is a delivery landing mid-count. Choose a date just after your normal delivery slot, or ask suppliers to hold shipments for that week and receive everything before you start. If a container is due in three days, move the count, not the container.
T-7 — build the count sheet from a fresh export. Export inventory for each location (Products → Inventory → Export, as of November 2026) and freeze the expected quantities with a timestamp in the sheet header. Counters get a blind copy with the expected column removed; the reconciler keeps the full copy. Build zones by physical location, never alphabetically — the sheet should follow the walk.
T-7 — decide the trading plan in writing. Two workable options: close for the day, or count after the last sale and post the same night. What doesn't work is counting all morning while the store sells, then posting at 4pm — every unit sold in between becomes an invented variance. Write down the cut-off time and tell the team.
T-3 — assign pairs and stop transfers. Two people per zone (one counts, one records) cuts both miscounts and shortcuts on high-value shelves. Freeze inter-location transfers for the count window: a transfer marked ready to ship commits stock at the origin, and a half-finished transfer on count day is a variance you created yourself.
The count sheet
Ten columns. Anything more slows the floor down; anything less and reconciliation becomes archaeology.
| Zone / bin | SKU | Variant | Expected (frozen) | Counted | Counter | Variance | Reason code | Recount | Posted |
|---|---|---|---|---|---|---|---|---|---|
| A2-1 | MUG-CEL-14 | Celadon mug 14oz | 96 | 91 | JT | −5 | DMG | No | Y |
| A2-3 | PLT-SNK-8 | Stoneware plate 8" | 142 | 142 | JT | 0 | — | — | — |
| B1-4 | BWL-RAM-S | Ramen bowl, small | 58 | 64 | AR | +6 | REC | Yes | Y |
| C3-2 | GFT-BOX-S | Gift box, small | 310 | 310 | AR | 0 | — | — | — |
Rules that make the sheet trustworthy: the expected number is frozen at export and never updated mid-count; reason codes are short (DMG damaged, REC receiving error, POS till or sync issue, RET return not restocked, UNK unexplained); a recounted row is flagged, never silently corrected; and the Posted column is signed by the person who applied the adjustment. A count nobody posted did not happen.
Reconciliation, with the arithmetic visible
Reconciliation is three moves, in order: compare, explain, then adjust. A ceramics store counting 320 SKUs on a Saturday, illustrative numbers:
- Line accuracy: 9 of 320 SKUs mismatched → 311 ÷ 320 = 97.2%.
- Unit variance: net −38 units across the catalog.
- Value: at a blended cost of $6.40, about $243 of recorded stock that did not exist (estimated).
- Concentration: 5 of the 9 mismatches sit on one shelf, which was re-slotted the previous Tuesday.
That last line is the whole point of the exercise. The number tells you the baseline; the concentration tells you what to fix. Five mismatches on a freshly re-slotted shelf is not theft and not shrinkage — it is a put-away process that didn't update bins, and the fix is a bin label and a rule, not a security camera.
Then: adjust each counted variant to the counted quantity with the reason Count (the adjustment history then records it as "Inventory manually counted", dated and attributed), use Damaged for anything physically unsellable, and leave the small unexplained tail marked as unresolved rather than filing it under Correction. Filing losses as corrections is how a real theft pattern stays invisible for a year.
One reference point worth having open while you reconcile: the stock reconciliation routine, which covers reconciling Shopify against a POS, a 3PL, or a spreadsheet as well as against the shelf.
After the count: three numbers to keep
Pull these into one page and file it with the valuation:
- Line accuracy, so next year's count has something to beat. The two versions of the formula — line-level and unit-level — disagree on purpose, and knowing which one you recorded matters more than the number itself (the accuracy formula).
- Value at cost, plus the variance in dollars, which is the figure that belongs in your books and in any insurance conversation.
- The top three causes by dollar impact, each with a named owner and a one-line process change. A count that produces a valuation and no process change is a very expensive stock list.
One downstream effect worth planning for: a stock take resets quantities, which means any velocity number you calculate across the count date is contaminated — units "sold" on count day may actually be a correction. Break velocity calculations at the count date, and if a SKU's history includes weeks at zero, use in-stock-only velocity rather than a calendar average, since the zeros were unavailable stock rather than absent demand.
Common questions
What is a stock take in Shopify? A full physical count of every SKU at every location, taken on a fixed date and posted back into Shopify as inventory adjustments so the system matches the shelf. Shopify's native count tool handles the scoping, counting, and review; the valuation and the variance investigation are still your work.
How often should a Shopify store do a stock take? Once a year for most small stores, plus after a system migration, a warehouse move, or before selling the business. Between stock takes, run cycle counts on a rotating slice so the annual count isn't your only source of truth.
Do I have to close the store to count? No, but you have to count when nothing is selling. Count after the last sale and apply the adjustments before trading resumes the next day. Counting during trading hours and posting later overwrites the units sold in between, which turns real sales into invented shrinkage.
Does a stock take update inventory in Shopify automatically? Applying a count writes adjustments that bring recorded quantities in line with what was counted, and each change lands in the variant's adjustment history. Values, damaged write-offs, and any variance you want itemized at cost need separate handling.
What's a realistic time for a 500-SKU stock take? One to two staff-days including preparation and reconciliation for a single location with two counters — roughly 2 minutes per SKU to count and a further hour or two to reconcile and post. Multi-location stores and 3PL stock add their own sessions, and those should be booked as separate days.
Next step
Put the date on the calendar this week — ideally the week after your next big delivery — and build the count sheet from a fresh export rather than last quarter's. If you want the counting mechanics in depth before then, the scope-and-apply workflow lives in Shopify inventory counts, and the cadence that makes the next stock take uneventful is in building a count cadence.