Omnichannel order management: the promise nobody watches

What omnichannel order management is

Omnichannel order management is the practice of running inventory, orders, fulfillment and returns as one pool across every channel a brand sells through, so an order can be placed in one channel, fulfilled from another and returned to a third. Verhoef, Kannan and Inman named that collapse of channel boundaries in 2015. What unification bought was one decision, not one owner: you get one place to choose where an order should come from, and no place that notices when it does not arrive. We see this most clearly in one configuration: a brand selling from its own stores alongside a marketplace of drop-ship suppliers. An order routes cleanly to a store, the pick comes up short, and the record still reads as fulfilled by the node that never fulfilled it. Routing worked. Nobody owned the order that stopped.

The store is a node in that, not a legacy channel. E-commerce was 16.9% of total US retail sales in the first quarter of 2026, which means most of the demand still sits in a building with a door.

Omnichannel versus multichannel order management

Multichannel means every channel keeps its own stock and its own orders. Omnichannel means one pool, one order record, and a routing decision that can send any order to any node. The difference nobody states is the blast radius: pool the stock and an error in the number stops being contained to one channel, it is wrong everywhere at once.

• Where stock lives: multichannel holds a separate allocation per channel, omnichannel holds one pooled number every channel sells against.
• Who fulfills an order: multichannel ships from the node that owns the channel, omnichannel picks a node per order, and it can be a store.
• What a customer can do: multichannel keeps buying, collecting and returning inside one channel, omnichannel lets those three happen in three.

What an omnichannel order management system actually does

Strip out the vendor language and the capability set is five things, in the sequence an order meets them.

• Unified inventory visibility: one number for what exists and where, across warehouses, stores and drop-ship partners. Everything below it depends on that number being right.
• Available to promise: the calculation that decides what the customer is told at checkout about when they will get it. This is the promise, in software, and it is computed once.
• Distributed order routing: the rules that choose which node fulfills each order, weighing stock, distance, cost and capacity.
• Flexible fulfillment: ship-from-store, pick up in store, curbside and split shipments, the point at which ecommerce fulfillment stops being one warehouse and becomes a network.
• Returns across channels: a return accepted in any channel and credited against the original order, wherever that order was placed.

Be strict about available to promise, because it is the only capability the customer ever sees. Gallino and Moreno found that sharing reliable inventory availability across channels changes what shoppers actually do. The word carrying the weight is reliable. An availability signal that is wrong changes behavior too, just not in the direction you wanted.

Every channel you add multiplies the ways a promise can break

Adding a channel does not add one more thing to manage. It adds a multiplier, because every new node is a new place an order can stop with nobody being told. At order level it looks like this.

• An oversell against a pooled stock number, found when the pick fails.
• A store pick that comes up short after the customer was told it was ready.
• A pick-up-in-store order sitting uncollected with nobody chasing it.
• A split shipment where one leg never scans and nobody reconciles the parts.
• A reroute to a second node that happens after the promised date was quoted.
• A return accepted in one channel and never credited against the order in another.

Two of those are worth taking separately. Pick up in store is an operations trade-off rather than a convenience feature, and Gao and Su modeled how it moves workload onto the store rather than away from it. Buy anywhere and return anywhere is the heaviest load omnichannel adds, and most coverage treats it as a policy checkbox. NRF and Happy Returns projected returns at 15.8% of retail sales in 2025, about $849.9 billion, and 19.3% of online sales. A delivery exception outbound and a missing RMA in shipping inbound are one failure seen from two ends.

Why the system that routed the order does not catch the break

First, the inventory record is not accurate enough to pool. Auditing roughly 370,000 inventory records at a large retailer, DeHoratius and Raman found 65% of them did not match what was actually on the shelf. Pooling stock does not fix a wrong number, it distributes it.

Second, the data moving between systems is wrong more often than anyone designs for. GS1 US and the Auburn lab found 69% of orders moving between brands and retail partners carried data errors without item-level EPC data. The standards answer is shared event data, not shared dashboards, and event data is what the GS1 EPCIS standard carries.

Third, an order management system is built to decide, not to watch. It computes available to promise once, at checkout, and never re-asks an hour later whether that date is still true. A single-channel order management system has the same blind spot, multiplied here by every node. It is not only a service matter either: under 16 CFR 435.2, a seller that learns it cannot ship on time must seek the buyer's consent to the delay and refund promptly if consent is not given. Learns is the operative word, and nothing in the routing stack is built to learn it.

How to tell whether your omnichannel order management is working

Five questions, asked of the operation rather than the software. A shrug is an answer.

• Time to detect: how long between an order stalling and a person knowing. A bad answer is a number nobody has measured.
• Who is told: the customer, or a dashboard nobody opens. A bad answer is an exception report emailed nightly.
• Inventory truth: when the pooled number was last reconciled against a physical count. A bad answer is a cycle count that slipped in November.
• Exception ownership: which team owns an order sitting between two nodes. A bad answer is the store thinking it is CX and CX thinking it is the store.
• Contact rate after unification: order-status contacts per thousand orders, before and after you pooled the channels. A bad answer is that it went up.

Measure contact rate across channels, not per channel, because shoppers do not stay in one: a study of 46,000 shoppers found 73% used more than one channel. Your order management software will answer the middle three. Question one is the one nothing in the routing stack is built to answer, which is why it is the one worth asking.

What has to sit on top: detect, decide, act

Something has to watch the promise after routing has made it, across every channel and node. Detect the break at order level: the oversell, the short pick, the uncollected pick-up-in-store order, the split shipment leg that never scanned. Decide the remedy: reallocate to the next stocked location, hold the second leg, give the customer a choice, or escalate. Act in the systems that own the record, before the customer finds out. We work backwards from the promise and flag every order that has not progressed to where it should have.

Plainly, what we are not. Keeyu is not an order management system. We do not hold inventory, compute available to promise or route orders to a node. We sit on top of whatever does. We are not a helpdesk, a chatbot or a returns portal either, and that is structural: a helpdesk replies to a customer who already knows something went wrong, so it cannot act first. Our category is proactive e-commerce operations, run as post-purchase operations rather than support. Every order is a promise. Keeyu keeps the promise.

Watch the promise, not just the routing

You unified the channels. One pool, one order record, one place that decides which node fulfills what. Nothing in that stack is watching whether the date it promised at checkout is still true an hour later, and every node you added is another place an order can stop quietly. Keeyu detects the break across your channels and nodes, decides the fix and acts on it before your customer has to ask. If orders are stalling between nodes and you are hearing about it from customers, book a demo.

• Verhoef, P. C., Kannan, P. K. and Inman, J. J. (2015). "From Multi-Channel Retailing to Omni-Channel Retailing: Introduction to the Special Issue on Multi-Channel Retailing." Journal of Retailing 91(2), 174-181. Journal of Retailing
• U.S. Census Bureau. Quarterly Retail E-Commerce Sales, First Quarter 2026. Census Bureau release
• Gallino, S. and Moreno, A. (2014). "Integration of Online and Offline Channels in Retail: The Impact of Sharing Reliable Inventory Availability Information." Management Science 60(6), 1434-1451. Management Science
• Gao, F. and Su, X. (2017). "Omnichannel Retail Operations with Buy-Online-and-Pick-up-in-Store." Management Science. Management Science
• DeHoratius, N. and Raman, A. Inventory Record Inaccuracy: An Empirical Analysis. Management Science, 54(4). doi:10.1287/mnsc.1070.0789
• GS1 US and Auburn University RFID Lab. EPC/RFID Retail Supply Chain Data Exchange Study, executive summary. GS1 US study summary
• GS1. EPCIS and Core Business Vocabulary standard. GS1 EPCIS
• National Retail Federation and Happy Returns. 2025 Retail Returns Landscape. NRF returns research
• Federal Trade Commission. 16 CFR 435.2, Mail, Internet, or Telephone Order Merchandise Rule, eCFR. eCFR 16 CFR 435.2
• Sopadjieva, E., Dholakia, U. M. and Benjamin, B. (2017). "A Study of 46,000 Shoppers Shows That Omnichannel Retailing Works." Harvard Business Review. Harvard Business Review