Inventory

How to Fix Inventory Truth Across Shopify and a 3PL

How to Fix Inventory Truth Across Shopify and a 3PL
By Ruchit Dalwadi · Reviewed by Lalith Nandan Kalava · · 16 min read

To fix inventory truth across Shopify and a 3PL, stop asking either system to be the stock record. Put one inventory ledger above both. It owns on-hand and allocations, publishes a separate available-to-sell number to Shopify and to wholesale, and checks itself against the 3PL on a fixed cadence.

Picture a Tuesday morning at a hypothetical brand that sells on Shopify, takes wholesale orders, and ships everything from one 3PL. The customer service inbox holds a run of messages that all say the same thing: order confirmed on Shopify on Friday night, card charged on Saturday, and now an email saying the style is short. The ops manager pulls the Shopify inventory report and the 3PL portal, and the two show different counts for the same SKU. A wholesale buyer calls about a PO that was due to ship yesterday and asks, politely, whether the units were ever there. Nobody in the room can answer with confidence. The numbers exist. They do not agree.

How do you fix inventory truth across Shopify and a 3PL?

The short answer to how to fix inventory truth across Shopify and 3PL operations is that you stop treating Shopify as the source of truth for stock. Shopify is a sales channel. The 3PL is an execution partner. Neither is designed to be the master record of what your brand owns, what is allocated to wholesale, and what is available to sell across both. That role belongs to a system in the middle that holds the canonical on-hand number and pushes available-to-sell figures out to each channel. The Inventory Truth Scorecard measures how far your own operation sits from that.

Across the customers we are onboarding right now, the route into this problem is nearly always the same. They started on Shopify, added a 3PL when in-house pick and pack stopped working, layered in wholesale through Faire, NuORDER, or direct EDI, and bolted on a sync app to keep the storefront and the warehouse in approximate agreement. It works for a while. Then a second channel starts drawing on the same units, and the gap between approximate and accurate becomes a refund on one side and a short-shipped PO on the other.

The fix runs in six steps, and the order is the point, because each step depends on the one before it.

  1. Name one system of record for on-hand by location. It is not Shopify and it is not the 3PL portal.
  2. Post allocations when an order or PO is committed, not when the warehouse picks it.
  3. Publish channel-aware available-to-sell to each channel instead of copying one number everywhere.
  4. Hold the 3PL to posting deadlines for receipts, shipments, and count variances.
  5. Run a four-layer reconciliation cadence so drift is caught the day it starts.
  6. Gate returns through QC before they count as sellable.

What does inventory truth actually mean?

Inventory truth, a term from Uphance’s 6 Breakpoints of Apparel Operations framework (Breakpoint 3), is the property of a single, trusted on-hand and available-to-sell number that every channel, team, and report agrees on at the same moment in time. It has three components: on-hand, the physical count of units in a location; allocated, the units already promised to an order, a wholesale PO, a transfer, or a hold; and available-to-sell, which is on-hand minus allocated, optionally adjusted for safety stock or channel reserves.

To reconcile inventory across Shopify and a 3PL is to make those three numbers agree in every system that shows them: the on-hand at the warehouse, the committed pool against open orders, and the available-to-sell on the storefront and the wholesale portal.

When inventory truth holds, every system derives its number from the same source, customer service can quote ship dates without guessing, and finance can close the month without three reconciliations. When it breaks, every team keeps its own private spreadsheet of what it thinks is real, and the ops manager spends Mondays adjudicating between them.

Why does Shopify and the 3PL drift apart?

The drift is not caused by bad software. It comes from an architecture in which two systems both act as the source of truth and were never designed to share a ledger. Shopify holds an inventory number per location and deducts when an order is placed. The 3PL holds a number per bin and deducts when a pick is confirmed. Between those two events there is a window. In it, the order can be cancelled, the warehouse can find the unit damaged, or the SKU can be short on the shelf even though the system said it was there.

Wholesale widens the window. A wholesale shipment gets picked at the 3PL and never touches Shopify. A factory receipt is advised as one ASN and arrives across several partial deliveries. A retailer cancels a PO after the pick list is generated. A DTC return sits in a QC bin for days before anyone posts it. Each event is a small drift, and across a full catalog in a peak window the drifts add up to oversells.

The pattern I notice repeatedly when I am in customer calls is that the reconciliation problem traces back to four configuration decisions, at least one of which was made by default instead of on purpose: which system owns on-hand, which channel each location maps to, how returns post, and when wholesale commitments deduct from DTC availability.

This is Breakpoint 3 of the 6 Breakpoints of Apparel Operations: inventory truth gets weaker. It rarely arrives alone. When production and supply execution drift from the plan (Breakpoint 2), receipts land late or short and the on-hand number absorbs the error. A weak on-hand number then feeds Breakpoint 4, because every order written against it inherits the doubt. The loop also runs backward: mis-picks and unposted returns in the warehouse (Breakpoint 5) corrupt the count from below, which is the subject of the 3PL blind spot.

What are the symptoms of broken inventory truth?

The symptoms are operational, not technical. Customer service triages oversells as a daily task instead of a rare event. The wholesale team holds units back by hand because it does not trust the available number, which makes DTC look short of stock it has, which then drives discounting on items that are not deep. Cycle counts uncover larger variances than expected, and the warehouse gets the blame even when the root cause is a stale feed from the storefront.

Finance feels it during close. Cost of goods sold cannot be tied to units shipped because the opening and closing unit counts do not reconcile to the activity in between. The CFO asks for one inventory roll-forward and gets three, from the warehouse manager, the e-commerce manager, and the wholesale lead. Each is internally defensible. None of them match.

The most expensive symptom never reaches a dashboard. Wholesale buyers stop trusting your ATS file, pad their orders, or shift volume to brands that ship complete. The account is not lost in one season. The relationship cools over several.

What does the actual cost look like at a $15M brand?

Here is the back-of-envelope for a $15M apparel brand running wholesale, DTC, and a 3PL. The first cost is revenue and reputation: a 2 to 3 percent oversell rate during peak weeks, which lands as refunds on the best-selling styles, customers who do not come back, and stockists shorted on the sizes that were selling. Apply that rate to your own peak-week order volume to see what it is worth to you.

The second cost is the count itself. Someone spends 6 to 9 hours a week reconciling inventory across Shopify, the 3PL portal, and the wholesale system, and the number they rebuild is the number the buy, the reorder, and the wholesale allocation are made from. When it is late or wrong, those decisions are too.

The third is one FTE effectively doing data plumbing across spreadsheets and CSV exports. The title on the role is operations manager or DTC manager, so the brand is paying for an analyst, a buyer, or an account manager it is not getting. A junior hire does not fix it, because the work is structural. The month-end version of the same work is laid out in a monthly 3PL-to-ERP reconciliation sequence.

Why is a sync app not enough?

A sync app keeps two endpoints in approximate numerical agreement. It does not own the concept of allocation. Take one SKU as a worked example. Assume the 3PL has 38 units in the bin. Assume 12 of them are reserved for a wholesale ship window that opens next Monday, 6 sit in a quality hold, and the remaining 20 are available to DTC. A sync app copies 38 to Shopify, because it treats inventory as a single number per SKU per location, which is the data model Shopify and most 3PL portals use. It breaks as soon as a unit can be claimed by more than one buyer, because the storefront, the 3PL portal, and the wholesale order book each hold a count that is locally consistent and globally incoherent. Where Shopify’s own inventory model holds and where it gives out is covered in Shopify inventory management for apparel.

Sync apps also assume the 3PL is the physical truth and Shopify is the channel display. That is half right. The 3PL knows what is in the bin. It does not know what should be in the bin, because it does not see open factory POs, in-transit ASNs, returns authorized but not yet received, or transfers between facilities. Without that context, the 3PL number is accurate only about the past. Available-to-sell is a forward-looking number, and you cannot derive it from a backward-looking source.

What does the architectural fix look like?

The fix is to designate a system in the middle, between the channels and the warehouse, that owns inventory as a record. That system holds the canonical on-hand by location, computes allocations against open DTC orders, open wholesale POs, transfers, and holds, and publishes available-to-sell out to each channel with channel-specific rules. Shopify gets a feed that excludes wholesale reservations. The wholesale portal gets a feed that excludes DTC safety stock. The 3PL receives outbound orders to pick and reports back actuals, but it is no longer asked to be the source of truth for what the brand owns.

In practice that means posting 3PL receipts to on-hand as they land, posting allocations the moment an order is committed, holding returns in a state that separates sellable from inspection-pending, and handling multi-location math when stock sits at more than one 3PL plus an in-house pick station for VIP orders or sample sends.

How do you handle wholesale and DTC sharing the same pool?

Here is the position I will defend without hedging: wholesale should not run through Shopify’s native flow. Shopify treats inventory as a single pool against a single channel. It has no native concept of wholesale-committed stock that is physically in the warehouse but not available to DTC.

The hardest case is a single SKU that wholesale and DTC both sell from the same physical pool. The temptation is to split the pool, give each channel a fixed share, and let each sell against its quota. That works on paper. In practice it leaves units stranded on one side while the other sells out.

The better approach is a shared pool with prioritized allocation, which is what channel-aware available-to-sell means. Wholesale POs reserve units when the order is confirmed and the ship window is set. Units for a planned drop are reserved when the drop is scheduled, not when it goes live. DTC sells against whatever remains, with a configurable safety buffer if you want to protect a launch or a top seller. When wholesale ships, the reservation converts to a deduction. When wholesale cancels or shorts, the units flow back into the available pool automatically. A sync layer built to copy one number cannot do this, because it has no model of a wholesale order, its ship window, or its shortages. Who gets a scarce unit is its own subject, covered in how to write an allocation rule.

Magnolia Pearl is the reference for this under pressure. The brand runs frequent product drops with same-day fulfillment, the conditions under which a shared pool oversells first, and its oversell rate held under 0.5 percent through peak.

How do you keep the 3PL feed honest?

The 3PL feed is honest when three things are true. Receipts are posted within a defined SLA, usually same-day for full container receipts and within twenty-four hours for parcel returns. Cycle counts are scheduled and the variances feed back into the master record with a clear adjustment reason. Outbound shipment confirmations are sent in near real time, not in a nightly batch that leaves the ledger counting units that left the building that morning.

If your 3PL cannot meet those three conditions, the inventory architecture cannot be accurate no matter what software sits in the middle. A software fix does not cure a 3PL SLA problem: the system in the middle can absorb operational noise, but it cannot manufacture data the 3PL does not send. The reverse also holds: a new 3PL does not repair a missing system of record, a point the 3PL vs in-house decision tree makes in more detail.

What is the right reconciliation cadence?

There are four layers, and each has a different job.

LayerWhen it runsWhat it catches
Order deductionThe moment an order or PO is committedThe same unit sold twice across channels
Delta sync from the 3PLEvery receipt, pick confirmation, and adjustment, posted hourly or fasterDrift inside the trading day
Full snapshot reconcileNightly, line by line against the 3PL inventory fileSilent warehouse adjustments and timing gaps across the day boundary
Cycle count auditWeekly, on the top-moving and highest-value SKUsPhysical errors that no file comparison can see

The trap in the first layer is that Shopify does it natively, but only for Shopify orders. It does not know that a wholesale PO confirmed a few minutes ago should also reduce the pool.

The second layer is where a once-a-day sync fails: during high-volume windows a daily feed guarantees drift. In the third, variances above a threshold you set, in units or in value, go to a person for review the next morning. The fourth is deliberately narrow. It covers the SKUs that would cost the most in oversells or stranded stock if they were wrong and leaves the long tail to a slower rotation, covered in choosing a cycle count cadence.

If you skip a layer, the layers above and below it absorb the variance until they cannot, and the Tuesday morning inbox comes back.

How do you handle returns without breaking the ledger?

A return is not on-hand when it arrives at the dock or when the parcel is opened. It is on-hand when QC clears it and it is put away. Brands that post returns on receipt oversell against units that turn out to be damaged. Brands that batch returns to month-end understate available-to-sell through the weeks when that stock would have sold at full price.

Returns should post to inventory in days, not weeks. The architecture that holds is a returns receiving event that creates a pending-restock unit, a QC event that approves or rejects it, and a putaway event that posts to available-to-sell the same business day QC clears it. Magnolia Pearl, which handles returns at scale through collection boxes on top of frequent drops, is the kind of operation where a slow returns flow shows up at once. If your returns sit for weeks, the brand is running on a stale number without knowing it. The wider failure is described in the reverse logistics loop.

What does multi-entity wholesale do to the reconciliation problem?

It multiplies it. When two or three brands or entities share one warehouse, the question becomes which entity owns which on-hand pool, how transfers between entities post, and what the B2B portal shows a buyer shopping across brands. A logged-in buyer should not see units that sit in the same building but are committed to another entity’s DTC pool. If the ledger does not track ownership at the unit level, the brand books PO lines the warehouse cannot fill, and the chargeback follows.

Lufema shows what holding that boundary is worth. It runs multi-entity wholesale distribution with a multi-brand catalog and a B2B portal, and its inventory accuracy reached about 99 percent, up from 90 to 95 percent.

What are the operational anti-patterns to avoid?

Four come up again and again, on top of treating the 3PL portal as the system of record.

The first is reconciling on a CSV export. If your weekly routine is to export from Shopify, export from the 3PL, paste both into a spreadsheet, and eyeball the variance, you are not reconciling. You are documenting drift after it has already cost you orders.

The second is using safety stock buffers as a substitute for channel-aware available-to-sell. A blanket buffer hides the symptom and strands inventory. It is a tax on every SKU to cover an architecture gap. The alternative to manual holds is described in resolving cross-channel inventory conflicts.

The third is letting marketing schedule promos without a hold on the inventory pool. If the email goes out before the reservation is in the ledger, you will oversell. The promo calendar and the reservation logic have to share a system. A Slack channel does not count.

The fourth is changing 3PLs to fix reconciliation. The variance comes from inventory living in several systems, so it follows the brand to the next building.

What this means for an apparel operations team

If your team spends Monday mornings reconciling Shopify against the 3PL portal, you do not have a sync problem. You have a source-of-truth problem. The fix is a record-keeping layer in the middle that holds on-hand, allocations, and available-to-sell as distinct fields that channels read from and do not write to.

The signal that you are ready for this change is usually not a dramatic incident. It is the slow accumulation of small ones: a handful of oversells in a week, a wholesale buyer who asks twice about ship-complete rates, a finance close that runs long because the inventory roll-forward did not tie. Together they are the cost of an architecture that held while the brand sold through one channel and stopped holding once wholesale, DTC, and a 3PL were all drawing on the same stock. For brands running that operating model, the break usually shows up in the $10M to $20M zone.

The brands that get this right run product development, product data, production, inventory, orders, warehouse execution, payments, accounting, and reporting in one connected system. The ones that do not run them across 3 to 5 tools plus spreadsheets, and they pay for it in refunds, chargebacks, and an operations lead whose week goes to rebuilding a count.

Fixing inventory truth removes the problem of not knowing what you own, which every other operational decision rests on. Reconciliation is a question of where the truth lives. Put it in one place, define the cadence, and the storefront and the warehouse stop arguing.

Inventory Truth Scorecard

How accurate is your inventory really?

Nine questions estimate where your operation sits on the inventory-truth curve and how much revenue is at risk. Takes about three minutes.

Frequently asked questions

Where this fits in the Uphance platform

R
Written by
Ruchit Dalwadi
Head of Product, Apparel Operations, Uphance

Ruchit writes about product strategy for apparel operations, covering how mid-market fashion brands use connected workflows to manage product development, inventory, orders, warehouse execution, and reporting. As Head of Product at Uphance, he shapes the roadmap that ties PLM, PIM, BOM management, allocation, fulfillment, and warehouse operations into one system. His articles dig into apparel-specific operational mechanics: tech packs, spec sheets, putaway, pick-pack, landed cost, and the data plumbing that makes inventory truth possible across multiple channels and locations. He focuses on the workflow-level questions that separate generic ERPs from systems built for how apparel brands actually run.

L
Reviewed by
Lalith Nandan Kalava
Senior Product Manager, Reporting and Operational Analytics, Uphance

Lalith writes about operational reporting and analytics for apparel brands, covering how connected data across inventory, orders, fulfillment, and warehouse execution translates into reporting that supports real decisions. As Senior Product Manager for Reporting and Operational Analytics at Uphance, he builds the dashboards and KPI work that let finance and operations teams stop arguing over numbers and start running the business. His articles cover landed cost, COGS reconciliation, month-end workflows, margin analytics, and the data hygiene patterns that determine whether reporting can actually be trusted at the executive level. He argues that reporting becomes political only when the operational layer underneath it is fragmented.