Shopify POS location setup: which SKUs live where, and the opening routine that keeps it honest

Short answer: a POS location is a real place with its own stock count — Shopify deducts every counter sale from the location the device is set to, and syncs that deduction to your online availability. Get the architecture right (one location per physical place, SKUs assigned deliberately, a primary location that reflects where your inventory actually lives) and POS and online stay truthful with zero effort. Get it wrong and every week starts with "the website says 6 but the shelf says 3."

Instead of a checklist, this page attacks a typical setup plan the way a skeptical ops person would — a pre-mortem. Here's the plan most merchants make, here's where it breaks, and here's what to build instead. (Day-to-day sync behavior and troubleshooting live in the POS inventory guide; this page owns the architecture.)

The plan under attack

A bike-parts shop adds a Saturday market stall to their existing garage-and-storefront operation. The plan, as usually conceived:

1. Keep the one existing location ("Main store") for everything.
2. Sell at the market from a POS device pointed at that same location.
3. Bring "a bit of everything" and reconcile later.

Now the attack — each failure below is a real, common outcome, not a hypothetical:

Attack 1: two physical places, one location. The market stall sells 30 units of a common cassette and some cables. Shopify deducts them from "Main store" — the location that also represents the shop's shelf and the online warehouse. Saturday night, online availability is wrong by exactly the market's sales: oversells risk on anything the market sold the last of, and a Sunday-morning manual correction that becomes a weekly ritual. The failure is architectural, not behavioral — no amount of care fixes two places sharing one count.

The fix: one location per physical place. Settings → Locations → Add location, "Market stall," with its own stock. At the market, the POS device is set to that location (the POS app lets staff switch the active location per device, and it should be checked at open — Shopify Help Center, "Locations" and "Shopify POS hardware settings," accessed October 2026). Stock that travels to the market moves there as a native transfer (the transfers guide owns that workflow), so the stall's count is real, the shop's count is real, and nothing needs reconciling.

Attack 2: "a bit of everything." Without SKU selection, the market carries slow movers nobody there wants, and the fast movers run out by noon. The pre-mortem question: what will this location actually sell? A market stall selling bike parts has a narrow plausible range — tubes, tires, cables, consumables. Those get transferred in quantities sized to the day; the 400 tail SKUs never leave the shop. Which SKUs live where is a decision made per location, written down, and revised quarterly — not an accident of what fit in the van.

Attack 3: the primary location choice made by default. Your primary location is Shopify's default for new inventory and the default location in order routing. Most merchants leave it as "wherever Shopify created it" — often the address from account signup. If stock physically lives in the garage and the primary is set to the retail storefront, every inbound PO's default destination is wrong and every routing edge case leans the wrong way. The fix: primary = where the majority of stock lives and ships from, checked once, changed rarely (Shopify Help Center, "Setting a primary location," accessed October 2026).

Attack 4: the reconciliation ritual itself. The plan's step 3 — "reconcile later" — is the tell. Any setup that requires a weekly manual reconciliation between POS and online is architecture shouting that it's wrong. Locations done right make reconciliation unnecessary; what remains is counting, which is a different job with a different cadence (the cycle count guide).

The architecture that survives the attack

The same bike shop, rebuilt:

• Three locations: "Garage warehouse" (primary — bulk stock, online fulfillment), "Shop front" (retail shelves), "Market stall" (created seasonally, or kept year-round for simplicity).
• SKU assignments: all 1,100 SKUs are stocked at the garage; ~180 fast-moving consumables also stocked at the shop front; ~40 of those travel to the market on stall days.
• Movements are transfers, not edits. Friday's market box is a transfer Garage → Market stall (30 tubes, 20 tires, 15 cable sets). Sunday night, unsold stock transfers back. The transfer record is the reconciliation.
• Opening routine, every selling day: confirm the device's active location; glance at the location's low or zero items (the stall's count of 30 tubes should read 30 at open); sell. Closing routine: sell, then transfer remnants back before the count gets stale.
• New SKUs: created with the primary location as their home by default — which is why Attack 3 matters. A SKU that only ever lives at the shop front still gets its stock added at the primary and transferred in, keeping the paper trail in one direction.

What POS locations can't fix

Honest limits as of October 2026: locations don't do demand planning — Shopify will track per-location counts but won't suggest what to stock at the stall based on what it sold last market season; that math stays with you (the transfers guide above covers the lead-time thinking a recurring transfer route deserves). Locations also don't stop staff selling the last unit twice across locations — that's a committed-stock question governed by your channel settings. And hardware doesn't make any of this true by itself: a scanner that speeds up the market stall's sales is worth having (the POS hardware guide covers the buying order), but architecture first, hardware second.

Common questions

How do I add a location for Shopify POS? In the admin, go to Settings → Locations → Add location, give it a name and address, and decide whether it stocks inventory and fulfills online orders. The location then appears in the POS app's location picker — set each device's active location at open so counter sales deduct from the right count (Shopify Help Center, "Locations," accessed October 2026).

Should my retail store and warehouse be separate Shopify locations? Yes if stock physically lives in two places — a location is Shopify's unit of truth for where units are, and one location can't represent two shelves. Set the primary location to where most stock lives and ships from, stock each place with what it actually sells, and move stock between them with native transfers so the trail is recorded.

What happens to online inventory when POS sells an item? The POS sale deducts the unit from the sale location's count, and the storefront's availability updates accordingly once the sync processes (normally quickly, but treat it as near-real-time, not instant). If the market stall and the shop share one location, the storefront's count reflects both — which is exactly why separate locations matter as of October 2026.

How do I move stock between Shopify POS locations? Use a native transfer: Products → Transfers, origin and destination, quantities, then receive at the destination. For a recurring route — weekly market stall, store-to-pop-up — the same transfer structure repeats, and unsold units transfer back at close. Marking the transfer complete before the boxes physically arrive creates sellable stock that doesn't exist yet; receive when it lands.

What is the primary location in Shopify and which should I pick? The primary location is the default destination for new inventory and a tie-breaker in order routing. Pick the place where the majority of your stock physically lives and from which most orders ship — for most POS merchants that's the warehouse or stockroom, not the retail counter. You can change it later, but getting it right at setup avoids a stream of small wrongnesses (Shopify Help Center, accessed October 2026).

Next step: open Settings → Locations and answer two questions honestly: does one location currently represent more than one physical place, and is the primary where stock actually lives? If either answer is no, fix the architecture this week — it takes an afternoon, and every week after it starts with numbers that are already true.