Spark Inventory Blog

A product business can show 800 units on hand and still stock out on its most important channel next week. The usual cause is not a lack of reporting. It is a failure to reconcile multichannel inventory into one usable inventory position before making purchasing decisions.

For an operations leader selling through a mix of Shopify, Amazon, wholesale, marketplaces, and multiple warehouses, reconciliation means more than checking that inventory totals look similar. You need to establish which units are physically available, which are already committed, which are moving between locations, and which incoming receipts can be counted on. Only then can a per-SKU forecast and buying plan produce a reorder recommendation you can act on.

Start with one definition of available inventory

Every channel and system uses inventory fields differently. A warehouse may report physical on-hand stock. A marketplace may show sellable inventory after its own reserve rules. Your ERP may include inventory that is allocated to open wholesale orders. If the team compares those numbers without a common definition, the result is often false confidence.

For replenishment, the most useful measure is usually inventory position:

`Available inventory position = on hand - committed - unavailable stock + incoming inventory`

The exact treatment of incoming inventory depends on timing. Inventory on an approved purchase order is not equivalent to inventory received at the dock. Inventory shipped by a supplier but not yet received carries a different level of risk. Keep those states separate, then use expected receipt dates when calculating days of supply and reorder timing.

Also define unavailable stock explicitly. This can include quarantined goods, damaged units, samples, inventory held for retail locations, or units under investigation after a count variance. Leaving these units in available stock may make the total look right while the buying decision is wrong.

Set a reconciliation cutoff before comparing systems

Multichannel reconciliation breaks down when teams compare data pulled at different times. Amazon may have updated after a morning order batch, while the warehouse export is from the prior evening and the wholesale order system includes orders entered an hour ago.

Choose a cutoff time for the review, such as midnight local warehouse time or the close of the prior business day. Every source should be interpreted as of that point. If live data is available, the system can refresh more frequently, but the operating rule still matters: compare like with like.

This is particularly important around high-volume periods. A difference caused by a six-hour timing gap is not necessarily a data error. Treat it as a timing exception first, then determine whether it will affect a reorder decision. A $20 difference in a low-volume accessory SKU may wait for the next cycle. A 200-unit gap in a fast-selling core SKU may require an immediate check.

Reconcile at the SKU and location level

Do not begin with a company-wide inventory total. Aggregate totals can balance while individual SKUs are badly wrong. Reconcile by SKU, warehouse, and inventory status, then roll the numbers up for management reporting.

A practical sequence is to compare physical on-hand inventory from the warehouse or ERP against the inventory reflected in each sales channel. Then account for open customer orders, transfer orders, purchase orders, and recent receipts. For each mismatch, identify whether the issue is a transaction delay, a mapping problem, or a true quantity variance.

SKU mapping deserves special attention. The same product can have a warehouse SKU, a Shopify variant SKU, an Amazon merchant SKU, and a wholesale item code. Bundles create another complication: selling one bundle may consume two or more component SKUs. If those relationships are not mapped correctly, channel sales can reduce availability in one system without reducing it in the system that drives purchasing.

Use a controlled item master as the reference. It should identify the canonical SKU, channel-specific identifiers, pack sizes, units of measure, and whether the item is a standalone sellable unit, a component, or a bundle. A case of 24 and a single unit cannot be reconciled by matching counts alone.

Match the transactions that create most variances

Most differences are traceable to a small set of transaction types. Instead of asking why a balance is wrong in general, work through the transaction history that could have changed it since the last clean reconciliation.

The common sources are:

• customer orders that were placed, canceled, partially fulfilled, or returned
• warehouse receipts that are pending, short, over, damaged, or posted late
• inventory transfers that have left one location but not arrived at another
• adjustments from cycle counts, write-offs, kitting, or returns processing
• channel listing and SKU mapping changes that prevent updates from reaching the right record

A good exception log records the SKU, location, quantity difference, source systems, likely cause, owner, and expected resolution date. The goal is not to produce a perfect report. The goal is to clear exceptions quickly enough that planning decisions use trustworthy inventory.

For example, suppose a warehouse shows 500 units on hand for a core SKU. Your marketplace shows 470 available units, and Shopify shows 500. The 30-unit difference may be a marketplace reserve, a delayed fulfillment update, or an unposted transfer. Until you know which, do not add those 30 units to the available pool for purchasing. Flag them as unavailable or uncertain in the buying plan.

Separate channel availability from the shared supply pool

Not every unit should be freely available to every channel. A wholesale commitment, retail allocation, or fulfillment reserve may be commercially necessary even if the inventory has not shipped. Reconciliation should reveal these allocations, not erase them.

First calculate the shared supply pool. Then apply the rules that determine how it can be used. Some businesses reserve inventory for wholesale purchase orders. Others protect a minimum quantity for a high-margin direct channel or distribute stock across warehouses to meet service expectations. There is no universal allocation rule, but there should be one documented rule per SKU class.

This is where channel-level reports often mislead teams. A channel can look healthy because it shows available inventory, while the business as a whole has insufficient supply after accounting for commitments elsewhere. Conversely, a channel can look short because a reserve has been applied, even though incoming inventory makes the overall position acceptable.

Your buying plan should use the shared supply picture plus policy-driven allocations. It should not simply add up whatever each channel displays as available.

Recalculate demand and replenishment after inventory is clean

Reconciliation is not the end of the process. It is the input to a better replenishment decision. Once inventory, commitments, and incoming supply are credible, evaluate each SKU against its per-SKU forecast, supplier lead time, order cycle, minimum order quantity, and safety stock policy.

A basic reorder calculation considers forecast demand during lead time, plus the inventory buffer you need, minus usable inventory and reliable incoming supply. But the details matter. If a supplier has inconsistent delivery performance, planning against the quoted lead time may understate stock risk. If demand has shifted by channel, a simple historical average may overstate or understate the next few weeks.

Look at days of supply by SKU and location. A SKU with 45 days of company-wide supply may still have only five days of supply in the warehouse that fulfills your fastest channel. In that case, a transfer may be the right action before a new purchase order. A transfer can protect availability faster, but it also creates handling cost and may leave the sending location exposed. The right choice depends on lead time, freight, service commitments, and the risk of a stockout.

Make reconciliation part of the operating cadence

The right cadence depends on order volume and inventory complexity. A business with modest volume and a single warehouse may reconcile critical SKUs weekly and complete a fuller review monthly. A business with multiple fulfillment locations, fast-moving products, and active marketplace sales may need daily exception review.

What should not vary is ownership. Someone needs to own source-data quality, someone needs to resolve warehouse variances, and someone needs authority to adjust purchasing assumptions. Without clear owners, the same mismatch appears in every planning meeting and gradually becomes accepted as normal.

Spark Inventory can bring supported channel connections and file imports into a single operational view, validate the inputs used for planning, and use the reconciled position to maintain stock-risk reports, per-SKU forecasts, and a buying plan. It can prepare draft purchase orders from those inputs, but the operator reviews, adjusts, and approves each decision before anything is committed.

The practical standard is simple: do not ask a buyer to make a reorder decision from a number they do not trust. Reconcile the inventory position first, keep uncertainty visible when it cannot be resolved immediately, and let each buying decision reflect the stock you can actually sell or deploy.