In 2025 the stuff sitting in the wrong place, or missing from the right one, cost the retail industry roughly $1.77 trillion. That is the IHL Group number for inventory distortion, the polite term for out-of-stocks and overstocks combined. It works out to about 6.2% of worldwide retail sales. Out-of-stocks drive about two thirds of it, overstocks the rest.
So here is a question worth sitting with: if inventory is the single most expensive thing to get wrong in commerce, why do so many order management conversations treat it as plumbing?
Commerce, stripped to the bone, is one act repeated a billion times a day. You move a unit of inventory from a warehouse to a customer, as fast and as cheaply as you can. Everything else (the checkout, the promo engine, the loyalty program) exists to make that transfer happen more often and at a better margin. The order management system is the engine that runs the transfer. And inventory is its beating heart. Get the heart wrong and it does not matter how pretty the rest of the machine looks.
Let’s take a look at what “getting it right” actually requires, because it is more specific than the vendor slides suggest.
The number nobody wants to talk about
Retail inventory accuracy in the US averages around 63%. Physical stores sit near 65%. More than half of retailers (58%) operate below 80% accuracy. World-class performers hit 95% and up, which tells you the gap is not a law of physics. It is a systems problem.
Think about what 63% accuracy means in practice. For roughly one item in three, the number your system shows is wrong. You are promising stock you do not have and hiding stock you do. Every downstream decision (can I promise this order, where do I ship it from, is this shipment even profitable) is built on a foundation that is wrong a third of the time.
Why this matters: the customer does not grade on a curve. A single stockout sends 69% of online shoppers to abandon the purchase outright, and 63% of those who abandon never come back even after you restock. Nine percent permanently switch retailers after one bad experience. You did not lose a sale. You lost a customer, and you paid for the acquisition twice.
Stating problems without proposing solutions is lame, so let’s get specific about what an inventory layer has to do. There are five jobs. Not fifteen. Five.
-
Be accurate. Trust the number.
-
Know where everything is. One source of truth across every location.
-
Forecast and allocate it. Right stock, right place, right channel.
-
Promise it. Commit to the customer with confidence.
-
See across your own and dropship stock. Sell more than you own.
Every one of these maps to a purpose-built capability, not an afterthought bolted onto an order table. Here is how Kibo handles each.
Job 1: Be accurate
The enemy of accuracy is not bad math. It is drift. Feeds go stale, a warehouse adjustment lands late, two systems disagree about the same SKU, and suddenly you are overselling.
Two things keep the number honest.
First, safety stock rules and thresholds. You withhold a configurable buffer from the Available quantity so the storefront never sells into a number you cannot actually fulfill. You set floor and excess thresholds and get low-stock and out-of-stock indicators before a shopper hits a wall. This is the difference between “Only 3 left” being a nudge and being a lie.
Second, the sync layer. Kibo gives you Refresh and Adjust APIs for real-time updates, bulk import files for the heavy lifting, and a managed Delta Export Feed that pushes recently changed records out to your WMS, marketplaces, and downstream systems. The point is not the API count. The point is that everyone reading inventory reads the same truth, and the truth updates as reality does.
Modern real-time tracking can push accuracy toward 99% and up. The revenue is not hiding in the last mile of your supply chain. It is hiding in the gap between 63% and 95%.
Job 2: Know where everything is
You can only sell what you can see. Sounds obvious, and yet siloed, location-blind inventory is exactly why so many retailers cannot turn their stores into fulfillment nodes.
Kibo tracks inventory at the UPC plus location level across warehouses, DCs, stores, and vendor sites. One authoritative view, not a pile of spreadsheets that agree by coincidence. That single view is what makes order routing and omnichannel fulfillment possible in the first place.
On top of that sits the Real-Time Inventory Service, a dedicated high-performance service that feeds live availability to the storefront. In-stock status on the product page, “Only 3 left” messaging, store-specific availability for buy-online-pickup-in-store. It is on by default, which is the correct default.
For the operators, the Supply and Demand UI is the modernized admin view: color-coded stock statuses, a consolidated modal that shows supply against unfulfilled demand at a glance, and drill-down into the actual shipments an allocation is tied to.
Why this matters: real-time cross-channel visibility gets cited as the number one driver of omnichannel revenue. It is linked to 20 to 30% lower inventory costs and up to 30% higher conversion. And it is table stakes now, not a differentiator: 77% of leading US retailers already offer BOPIS, and 75% of those pickup customers buy something extra while they are in the store. That last stat is the whole game. Visibility does not just move the unit you already sold. It sells another one.
Job 3: Forecast and allocate it
Here is the trap. You can have the right total quantity and still stock out, because the right quantity is in the wrong place or promised to the wrong channel. Aggregate inventory is a vanity metric. Allocation is the real one.
Three capabilities do the work.
Inventory segmentation lets you ring-fence portions of stock for specific channels, customer groups, or fulfillment types using tags, and allocate by percentage or by discrete quantity. Reserve some for your highest-value B2B accounts, wall off a chunk for retail, keep the marketplace from cannibalizing store fulfillment.
Granular tracking lets you allocate by SKU, lot code, expiry date, serial number, and condition. That means you can run first-expired-first-out for perishables and lot-tracked goods, and treat “new” and “refurbished” as the different things they actually are.
Reservations hold inventory for items sitting in a shopper’s cart for a set window, or park stock for a specific B2B account. Nobody enjoys watching a unit vanish from their cart at checkout.
Why this matters: AI demand forecasting can cut forecast error by 30 to 50% (McKinsey’s range) and prevent up to 65% of stockouts. But a forecast is just an opinion until the platform can act on it. A brilliant prediction with blunt allocation controls is a very well-informed way to still put stock in the wrong place. Good luck with that.
Job 4: Promise it
A vague promise costs you the sale. A broken promise costs you the customer (see the 63% who never come back). So the promise itself is a feature, and inventory is what makes it credible.
Future Inventory, which is Kibo’s Available-to-Promise capability, lets you accept and allocate orders against inbound stock with confirmed delivery dates. Pre-orders, backorders against a real replenishment date, ATP that reflects what is actually coming and when. You configure it at the site and product level with limits on how far into the future you will promise. There is a calendar view so a human can actually reason about it.
Then there is the Assume Infinite Inventory flag, a per-record switch that lets you allocate beyond available levels for the cases where that is correct: dropship and made-to-order items where a “0 on hand” number is meaningless and a false stockout just throws away revenue.
Why this matters: a specific delivery date lifts conversion by 13 to 25% versus a fuzzy shipping range, and showing delivery dates is linked to a 36% drop in cart abandonment. More than 85% of shoppers now expect an ETA at checkout. ATP is the moment your accurate inventory stops being an operational hygiene metric and starts being a conversion lever. That only works if the number underneath is real, which is why Jobs 1 and 2 come first. There is no free lunch.
Job 5: See across own and dropship
Shoppers want the endless aisle. Your CFO wants less capital tied up in stock. These two are not friends, and dropship is the treaty between them.
The dropshipping market ran somewhere around $400 billion in 2025 and is growing in the low-to-mid 20s percent annually, depending on whose report you trust (they disagree, loudly). Over a quarter of online retailers now use dropship as a primary fulfillment method. The appeal is simple: expand assortment and revenue without carrying the distortion risk on your own balance sheet.
The catch is that dropship only works if you can see supplier stock in the same view as your own. Kibo’s unified multi-location model extends to vendor and dropship sites, so owned and supplier inventory live in one picture rather than two systems you manually reconcile at 2am. Assume Infinite Inventory handles the records where supplier availability is effectively unbounded, and the Delta Export Feed keeps partners synced as records change.
Sell what you own. Sell what your suppliers own. Show the customer one shelf. Do not make them care where the box ships from, because they don’t.
The point
Ten features, but really one system doing five jobs. Accurate data feeds a single source of truth, which feeds smart allocation, which makes a confident promise possible, which lets you extend the shelf beyond what you own. Pull any link out of that chain and the whole thing slows down, which in commerce is the same as bleeding money.
That is the part the “OMS as plumbing” crowd misses. The inventory layer is not the boring foundation you build the interesting stuff on top of. It is the interesting stuff. It is the difference between a $1.77 trillion tax and a 95% accurate, fast, promise-keeping machine.
Commerce is a race to move inventory from the warehouse to the customer. Win the inventory layer and you win the race. Lose it, and, well, the things you own end up owning you.
