Bottom line up front: EDI 846 inventory advice does not report how much stock is in your warehouse. It reports how much stock a retailer can sell right now. Between those two numbers sit inbound shipments, reservations, quality holds, and damaged goods, and everything in that gap turns into oversells, stockouts, and chargebacks. I learned that the hard way in March 2025, helping a home goods brand go live with 846, and I have seen the same failure three times since.
The project ran out of a 3PL warehouse in Ontario, California. The brand handed us inventory sync for more than 40 SKUs, and the retailer took orders against the numbers we sent. On day three, customer service blew up: our 846 showed 1,200 units available for a bestselling storage cabinet, the retailer accepted 900 orders, and the warehouse could actually ship only 640. Of the missing 260, about 180 were already reserved for other channels and 80 were sitting in a quality hold that nobody had released. The brand apologized to the retailer, canceled 260 orders, and the chargeback notice arrived the next day. The system had the right numbers all along. The feed sent the wrong ones.
What EDI 846 Inventory Advice Actually Reports: Four Quantities First
The 846 is the X12 Inventory Inquiry/Advice transaction set. The format itself is simple: the BIA segment sets the report type and date, LIN identifies the item, QTY carries the quantities, and SDQ pins them to a location. The format is never the hard part. The hard part is the definition of quantity, because one item has at least four different numbers inside a WMS:
| Quantity | How it is calculated | Can the retailer sell against it? |
|---|---|---|
| On hand | Physical units sitting on the shelf | No — it includes holds and damage |
| Available to sell | On hand minus reservations, quality holds, and damage | Yes — this is the number 846 should carry |
| On order | Shipped by the supplier, not yet received | No, but it tells the buyer replenishment is coming |
| Committed | Ordered already, not yet shipped | No — it has already been sold once |
Item one on my pre-launch checklist is getting all four definitions written into the retailer's implementation guide, and getting the brand, the 3PL, and the retailer to sign the same page. Most failures I have seen start here: the 3PL's WMS reports on-hand as available by default, the brand assumes the feed means available, the retailer assumes the same, and when the numbers disagree everyone blames everyone else. My rule is that the QTY segment carries available-to-sell as the single headline number, with on-order and committed reported under separate quantity codes, never folded into the headline. That is the entire job of an EDI 846 inventory advice warehouse feed: one honest number, reported the same way every time. Reporting a smaller honest number is fine. Reporting a fuzzy one is a disaster.
Here is the counterintuitive part: an 846 is a snapshot, not a promise. The second the file goes out, the numbers are already old — somebody in the warehouse may pick 50 units in the next minute. So never promise anything inside the document. Put promises in the contract; let the document state facts. Once you draw that line clearly, most of the arguments disappear.
How Often to Send EDI 846 Inventory Advice: Faster Is Not Always Better
Everyone's first instinct on frequency is that faster is better, ideally real time. My experience says the opposite: frequency has to match how fast your inventory actually moves and how fast your WMS refreshes. An EDI 846 inventory advice warehouse feed that runs faster than its own WMS refresh is just dressing up old numbers as new ones.
In October 2025, during peak season, I worked with a beauty brand that insisted on hourly 846 sends because a hero SKU could sell 300 units a day. But their WMS only refreshed its inventory snapshot in batches every 40 minutes, so the hourly feed often picked up the previous batch's numbers. The worst case: a face cream sold out at 11 a.m., and the noon 846 still showed 200 units available. The retailer took the orders, and the afternoon was spent canceling them in bulk. We changed the schedule to four sends a day, each one timed ten minutes after the WMS batch finished, and the oversells stopped. Frequency went down. Accuracy went up.
My tiered approach: slow-moving regular items get one or two full sends a day. Peak-season bestsellers get four to six sends a day, timed to land after the WMS refresh. Only drop-ship programs, where the retailer takes orders live against your 846, justify near-real-time pushes — and those must carry a snapshot timestamp so the receiver knows exactly when the number was true. That timestamp looks trivial until you are reconciling a dispute, and then it is the only thing that settles whose number was wrong.
The Cost of Reporting Too Much or Too Little
Over-reporting and under-reporting hurt in different ways, and both cost money. Over-report and you oversell: orders come in that you cannot ship, cancellation rates climb, and the retailer answers with chargebacks, demoted listings, and, in the worst cases, delisting. Under-report and you stock out on paper while the goods sit in your warehouse: lost sales, longer replenishment cycles, and rankings that slide while you fix nothing.
Run the math on one SKU (illustrative example): 8,000 units a month at $25 each. If the 846 overstates availability by 5% for a month, that is roughly 400 extra orders, and if half of them cancel, 200 cancellations at $8 each in chargebacks and service cost is $1,600 in direct losses — before counting the traffic lost to a demoted listing. If the feed understates by 5% instead, that is 400 units not sold, and at a 30% margin, $3,000 of gross profit gone in a month. A five-point definition error costs thousands a month and tens of thousands a year. I run this arithmetic with every client before go-live, and after they see it, nobody calls the quantity definition a detail anymore.
Pre-Launch Checklist: Seven Items, No Shortcuts
I run the same pre-launch checklist on every 846 project, and I do not go live with any item missing. One, the four quantity definitions are written into the implementation guide and signed by the brand, the 3PL, and the retailer. Two, QTY carries available-to-sell only, with on-order and committed reported separately. Three, every send carries a snapshot timestamp, taken after the WMS refresh. Four, full and delta sends are separated: one full reconciliation a day, deltas in between, so the full file is always there to fall back on. Five, zero-stock and discontinued SKUs are still reported — reporting zero beats not reporting, because silence looks like a missing file. Six, the low-stock alert and recovery thresholds are written down in numbers: below how many units you warn, above how many you restore availability, with no human judgment in the loop. Seven, there is a manual kill switch: if the numbers look wrong, someone can stop the automated feed within ten minutes and switch to manual reconciliation instead of letting bad numbers keep flowing out.
One last thing I tell every client: an EDI 846 inventory advice warehouse feed is not a technical project, it is a definitions project. The mapping takes half a day to build. Getting the definitions wrong takes three months of firefighting to undo. Nail down the four numbers, one frequency, and one checklist before you go live, and you will never have to write a hundred apology emails after.
By Lin Xiao, WMS Implementation Consultant







