Inventory accuracy at a 3PL is how closely the count in the warehouse system matches what's on the shelf, SKU by SKU and bin by bin. It stays honest when every unit is scanned at receiving and again at pick, cycle counts recount a slice of the stock on a schedule, and someone reads the variance report. Shrink is whatever's still missing after the recounts, and your agreement should say how it's measured, over what period and against which count.
What inventory accuracy means, and why the basis matters
A 3PL can quote you an accuracy figure three ways, and they don't agree. By unit, it's the share of units counted that matched the system. By location, it's the share of bins where the count matched to the unit. A value figure takes the dollar gap as a share of the dollars counted. Each one is honest, and each one answers a different question.
Say a quarter's cycle counts cover 500 bins holding 20,000 units, and 15 bins are off by a total of 60 units: that's 97% by location and 99.7% by unit, from the same count. The unit figure looks better because most bins are fine and the misses are small. The location figure tells you more about oversells, because a bin that's off by two is a SKU your store might sell when it's gone. When someone gives you one number, ask which basis it uses and over which months.
Inventory accuracy also isn't the same as order accuracy. Pick and pack error measures whether the right units went into the right box. A building can pick well and still drift on the shelf, through receiving miscounts, returns put back without a scan, or damaged units nobody moved out of stock. You want both numbers, and you want to know how the warehouse counts each one.
Where the count goes wrong between the dock and the box
Most variances start at a handful of predictable points. If you know them, you can read a variance report and guess the cause before anyone tells you.
- Receiving: an inbound gets counted by the carton, a carton holds fewer units than the label says, and the system books the label. Your packing list is the first count of record, so a short carton found later looks like shrink.
- Look-alike variants: two sizes or shades share a barcode, or one has none, so a scan can't tell them apart. GS1 US says each variation of each product you sell needs its own barcode (checked November 2026), which for a shirt in three sizes and three colors means nine.
- Putaway: someone scans units into the wrong bin, or puts them in the right bin without a scan. The total stays right while the location is wrong, so a picker can't find stock the system says is there.
- Picking: a picker grabs the neighbor SKU and the order goes out with the wrong item. The customer gets a size up, and the shelf is now one over on one SKU and one short on the next.
- Returns: someone opens and grades a parcel and puts it back on the shelf without a scan, or it sits in a returns area nobody counts.
- Damage: a crushed carton or a leaking bottle stays in sellable stock until someone tries to pick it.
Your store has its own version of the same problem. Shopify splits on-hand inventory into available, committed and unavailable (checked November 2026), where committed means units in unfulfilled orders and unavailable covers stock set aside as damaged, held for quality control, or kept as safety stock. If your 3PL holds a damaged unit as sellable and Shopify shows it as available, you'll sell it. So ask where damaged and quarantined units sit in the building, whether that area gets counted, and whether those units reach your store as unavailable.
How cycle counts work, and how often each SKU gets counted
A cycle count recounts a slice of the inventory every day or week, so every location gets counted over a cycle without shutting the building for a full physical count. It's how a warehouse catches a variance while the receipts and picks that caused it are still recent enough to trace.
How often a SKU comes up should depend on what a miss costs. Fast movers and high-value SKUs get counted more often, because they're the ones that oversell and the ones that hurt when a unit walks. Slow, cheap SKUs come up less often. Good programs also count on a trigger: when a bin hits zero, when a picker can't find a unit the system says is there, or when a return goes back to stock. Ask your 3PL which SKUs it counts most often, and why those.
Two habits separate a count you can trust from one you can't. The first is the blind count, where the counter sees the SKU and the bin and has to write down what's there, without seeing what the system expects. A counter who sees the expected number tends to find it. The second is the recount: when the number doesn't match, a second person counts again before anything gets adjusted, and only a variance that survives both counts goes into the system.
A full physical count still has its place. Your books need one too. IRS Publication 538 says a business keeping book inventory must take a physical inventory at reasonable intervals and adjust the book amount to agree with it (checked November 2026). Ask your 3PL for a dated count by SKU at your year end, signed off by both sides, and give it to your accountant.
How to read a variance report
A variance report lists the locations where the count didn't match the system, what the counter found, and what the warehouse did about it. If your 3PL can't send you one, ask why. If it can, these are the fields that make it useful.
| Field | What it tells you | What to ask if it's missing |
|---|---|---|
| SKU and bin, with the variant | Where the miss is, down to the size and color | Can I see variances by bin, so I can tell whether they cluster in one aisle? |
| System count and physical count | How big the gap was before anyone adjusted it | Was this a blind count, where the counter didn't see the system number? |
| Recount result | Whether a second person found the same number | Who recounts a variance, and does that happen before the adjustment posts? |
| Cause | Receiving miscount, wrong bin, wrong pick, unscanned return or damage | How many of last quarter's variances closed with no cause found? |
| Gross and net variance | Gross adds up every miss; net lets overages cancel shorts | Which one does the shrink clause in my agreement use? |
| Value | What the gap is worth at an agreed unit value | Which value applies: my landed cost, wholesale price or retail price? |
| Movement since the last count | The receipts and picks that touched that bin since its last count, returns included | Can I see the movement history for a SKU that came up short? |
Gross and net are where look-alike variants show up. Say the counts find 40 units short on a medium and 38 units over on the large next to it, so the net variance is 2 units and the gross is 78. Net makes that look like a good month. Gross tells you the shelf holds the right total in the wrong sizes, and your store's been selling mediums it can't ship. A pattern like that points at picking or putaway between two bins, and it's fixed with separate bins and a scan on both.
What the agreement should say about shrink
Shrink is the inventory that's gone after the recounts and the cause codes: units the warehouse received and can't account for. A swapped size isn't shrink, and neither is a unit sitting in the wrong bin. The agreement decides how anyone measures shrink and what happens when there's some, and the details matter more than any headline figure. The guide to 3PL contract terms covers the clause in context with the rest of the agreement; these are the questions specific to the count.
- Which count is the starting point: my packing list, your receiving count, or the first cycle count? How long do I have to dispute a receiving count?
- Is shrink measured gross or net, by unit or by value, and over what period?
- Are variances adjusted only after a blind recount, and do I see each adjustment with its cause?
- How is a missing unit valued, and at whose cost: my landed cost, wholesale or retail?
- Where do damaged and quarantined units sit, are they counted, and do they sync to my store as unavailable?
- Are returns scanned before they go back to stock, and is the returns area included in counts?
- Can I ask for a full physical count by SKU, at year end, before peak, or when I leave, and what does it cost?
- Can I see live inventory by SKU and the adjustment history whenever I want, without asking for a report?
Read the answers against your own data. If your packing list is the starting point, your supplier's carton counts need to be right, and that's worth a spot check of your own on the next inbound. How to send inventory to a 3PL covers the packing list and the carton labels that make a receiving count easy to check.
Checks you can run from your side each month
You don't have to wait for the warehouse to tell you something's off. Five checks cover most of it, and you can run them from two exports and a spreadsheet.
- Export on-hand inventory by SKU from the 3PL dashboard and from your store on the same morning, before orders start flowing. Compare the warehouse number to your store's on-hand figure, available plus committed plus unavailable, and list every SKU that differs.
- For each SKU that differs, look at its last receipt, its last return and its last adjustment. A gap that starts on a receiving date is a receiving question, and one that grows a little every week is usually picking or returns.
- Check every inbound from the last month against your packing list. If the receiving count was lower, ask whether anyone opened and recounted the short carton before the system booked it.
- Read the variance report for clusters: the same two SKUs, the same aisle, the same week. One cluster is a fix, and the same cluster three months running means nobody's fixing the process.
- Add up your support tickets for wrong items and items not received, and set them next to the variances on those SKUs. If customers are getting the wrong size, you'll often see it on the shelf as well.
If one SKU keeps drifting, ask for a trigger count on it every time a picker empties its bin, and look again next month. If your store and the warehouse disagree on many SKUs at once, it's often the integration, a sync that stopped or a location that got added to the wrong place, and that's worth checking before anyone recounts a bin. The Shopify fulfillment page explains how our counts reach your store.
How inventory is counted and checked at YNK
Inventory comes to 465 Mola Blvd, Dock 10, Elmwood Park, NJ 07407, and every unit gets scanned as it goes on the shelf, so look-alike sizes and shades never share a bin. If a product arrives without a scannable barcode, we label it before it goes to stock, billed as special-project labor at an estimated $35 an hour, so the pick scan has something to check against. Receiving is billed by the hour, and a labeled shipment with a packing list takes fewer hours than a mixed pallet without one.
Before anything goes in a box, someone scans it against the order, so a six-unit order gets six scans. That's how our pick and pack error stays under 0.3% across more than 1.2 million orders a year. Between receiving and pick, cycle counts reconcile the shelf to the system and catch variances early, so the number on your dashboard is the number on the shelf. Supplements, gummies, beverage and food-grade goods are tracked by lot and expiry date, and we pick the oldest sellable lot first. Returns are received and inspected, then restocked or disposed of by your policy.
You see the same live inventory we see, by SKU, in the dashboard, and Shopify, Amazon and TikTok Shop orders all draw from one inventory pool, so there's one count to check. An account manager assigned to your brand runs a weekly review of order flow, SLA and inventory with you. How loss and damage are handled is set in your agreement, sized to your volume and category, and we walk you through it before you sign. The answers are also in the inventory and accountability section of our FAQ, and the rest of the service is on the 3PL services page.
If you'd like your catalog priced, the 3PL intake takes about ten minutes and asks for your SKU list with sizes and weights. Someone from the team replies within one business day, and the proposal follows within five.
Questions founders ask about 3PL inventory accuracy
What inventory accuracy should I expect from a 3PL?
Ask for the figure with its basis, by unit, by bin or by value, over the last three months, and for the variance report behind it. A figure on one basis can't be compared with a figure on another, and on the same count by-unit accuracy reads higher than by-bin.
What's the difference between a variance and shrink?
A variance is any location where the count didn't match the system. Shrink is what's still unaccounted for after the recounts and the causes: units received and gone. A size swapped with its neighbor is a variance, and it isn't shrink, because both units are still in the building.
How often should a 3PL cycle count my inventory?
Often enough that every location gets counted over a cycle, with fast and high-value SKUs counted more often than slow ones, plus trigger counts when a bin hits zero or a picker comes up short. Ask which SKUs come up most and why.
How do I check inventory accuracy at YNK?
You see the same live inventory we do, by SKU, in the dashboard. Receiving is scanned onto the shelf, picks are scanned against the order, and cycle counts reconcile the shelf to the system. Pick and pack error runs under 0.3%, and the weekly review covers inventory.