Multichannel inventory: which channel may promise each unit

Which channel may promise the unit

A D2C brand with its own shop and one or more marketplaces has one pile of stock, and each listing is a promise that a unit will ship. Multichannel inventory is the rule for which channel gets that promise. If the shop and a marketplace can both sell the last unit, you cancel a buyer or you leave the unit unavailable to the channel that could have sold it.

Apply the rules below in whatever system already holds the count, including a spreadsheet. Nothing here describes how Shopify updates a quantity. Do not skip the buffer because a sync tool says the update is instant.

What a missed promise costs

Keep each marketplace's rule on the site it came from.

On sellercentral.amazon.com, a forum staff reply defines the cancellation rate as all seller-cancelled orders as a percentage of total orders in a given 7-day period. The same reply says the metric only applies to seller-fulfilled orders, and is not applicable to FBA opt-in sellers. FBA means Fulfilled by Amazon. A cancellation rate above 2.5% may result in deactivation of seller-fulfilled offers, and Amazon may require a Plan of Action before that deactivation.

Another Seller Central thread says selling partners are required to keep the pre-fulfilment cancel rate under 2.5%. The two threads disagree, so until the current Amazon.com help page is in front of you, use the staff reply's hedged wording rather than this hard requirement. Do not copy 2.5% onto a UK account until you have read the UK page.

A separate Seller Central thread says the cancellations that count include those the selling partner initiates, plus orders Amazon cancels automatically because the seller did not confirm shipment. Cancellations the customer requests through their own Amazon account do not count. An unconfirmed shipment can therefore move the rate without anyone choosing cancel. El from Amazon says some percentage of stock-outs may be unavoidable in normal business practice, and that Amazon still expects sellers to minimise those issues.

On eBay.com, the order cancellation policy states that cancelling an order because the item is out of stock results in a transaction defect. Sellers are responsible for fulfilling items they have sold, and they must give the accurate reason. Using an incorrect reason to avoid a defect, or to bypass eBay policies, is not allowed. A seller cannot cancel once the order has shipped, the buyer has reported non-arrival, a return is open, or more than 30 days have passed. This is the US eBay.com page, so do not copy it onto a UK account until you have read the UK page.

The refund, the labour and the account cost do not combine into one figure.

The refund is the amount you return for the order you cannot ship. In this illustration the buyer paid £40, so the refund is £40. The figure is invented, so the other costs have a number beside them. It is not an Amazon fee, an eBay fee, or an average from either site.

Labour is the time to cancel, tell the buyer, and correct the count if a unit was already picked. In the same illustration that takes 15 minutes. Those minutes are not part of the £40.

The account cost is separate, and these sources do not price it. On Amazon.com it is movement in the seller-fulfilled cancel rate over a given 7-day period. On eBay.com it is the transaction defect on an out-of-stock cancellation. Leave both as account events rather than adding them to the £40.

Count what you can still ship

Publish a count you can still dispatch.

1. Start from the last warehouse count you will defend for that SKU.
2. Subtract units already sold and not yet dispatched, on the shop and on every marketplace.
3. Subtract units reserved for bundles you are still offering. A live set has already promised its components.
4. Subtract a delay buffer: units other channels might sell before your quantity update reaches them. Set it from your sales pace and the longest gap you have seen between a sale and the other channels showing the new quantity. If you have not measured that gap, use the longest stretch someone might leave the update, then measure the gap before you shrink the buffer. The gap is your assumption, not a Shopify interval.
5. What remains is the shippable count. Only this count is split across channels.

If step 1 is wrong, syncing the result publishes the error in more places.

Three ways to split that count

Use the Amazon.com and eBay.com rules for one decision: whether a thin channel you might fail to fulfil should be capped or ring-fenced. A channel you can ship and confirm should still be allowed to sell.

Offer the full shippable count on a channel when the delay buffer already covers overlapping sales and a cancellation there is a refund plus labour, not an account event. The shop can take that full count if it has no account penalty of its own. A thin Amazon.com seller-fulfilled listing or an eBay.com listing cannot, because a miss there is an account event. At a low weekly volume, one or two seller cancellations put the 7-day rate on or over 2.5%.

Cap the channel when you want the listing to stay up. Publish a fixed maximum on the seller-fulfilled marketplace, below the shippable count, sized to orders you can pick, pack and confirm. Leave units you do not assign unpublished until you know which channel would treat a cancellation as an account event. These US rules do not rank margins, so size the cap to what you can fulfil.

Ring-fence when sharing the pile is the larger risk. Move a defined quantity to FBA, or to any location the other channels cannot sell, and take those units out of the shared count. The staff reply says the cancel-rate metric is not applicable to FBA opt-in sellers, so an FBA stock problem does not move the seller-fulfilled rate. If FBA sells slowly, those units sit while the shop runs short. That stranded stock is what you pay for the ring-fence. Accept it when it is cheaper than another seller-fulfilled cancellation on a thin listing. Other FBA performance rules are outside these sources.

How few cancellations 40 orders allow

These order counts are invented. The definition is the staff reply's: seller-cancelled orders as a percentage of total orders in a 7-day period, and seller-fulfilled orders only. Assume a cancelled order is in the total as well as in the numerator.

At 40 seller-fulfilled orders in the window, 2.5% is one order. One seller cancellation is 1 out of 40, equal to 2.5%. The consequence in the staff reply applies above 2.5%, so this single cancellation sits on the line, not over it. Two seller cancellations are 2 out of 40, which is 5%. That is above the line that may lead to deactivation of seller-fulfilled offers.

At 200 orders in the same window, 2.5% is five cancellations, and a sixth is 3%. The thinner channel can take fewer misses. Cap it or ring-fence it while you are still finding out whether it will sell.

If the same 40 orders were FBA, the staff reply leaves them outside this metric. The £40 refund and the 15 minutes illustrate a seller-fulfilled order you cannot ship yourself. The reply does not say an FBA unit that Amazon holds becomes that refund or that labour.

If the window is already 1 cancellation out of 40, the rate is already 2.5%. Cancel one more order and, if that order is included in the total, the rate is 2 out of 41, or 4.9% to one decimal place. Further orders lower the rate only when you fulfil them. One cancellation among 80 orders is 1.25%.

What a quantity sync leaves open

A quantity update can correct a channel that still shows a quantity you have already reduced, after the update lands. Four failures stay open.

Two buyers can both pass a quantity check on the last unit if neither order has reduced the published count. The delay buffer is there to absorb that overlap. Treat it as an operating risk: this article has no measurement of when Shopify, or any other shop platform, commits stock. With no buffer, the order you cannot ship becomes a cancellation you initiate. On the US rules above, that cancellation can count.

A bad warehouse count survives the sync. You have put the same wrong number on more listings. Recount before you raise a quantity after a stock-out.

Shipment confirmation is separate from the quantity sync. The Seller Central thread says Amazon cancels automatically when the seller does not confirm shipment, and that those cancellations count, so someone still has to confirm shipment after the quantity has updated.

Demand you leave running will keep creating orders. Updating a quantity does not end a listing, or stop an advert or email that still sends a buyer to a unit you cannot ship. A community post quotes guidance to end the listing or update the quantity as soon as stock changes, and that post is not the seller standards page. On a channel you control, stop the promise yourself rather than waiting for the next scheduled sync.

Priority, bundles, and when you pause

Put the priority on the SKU.

On a shared pile, unpublished residual units are safer than a guess. Assign the residual shippable count to a channel only when a cancellation there is not an Amazon.com cancel-rate event and not an eBay.com out-of-stock defect. Your shop can take that residual if the shop has no account penalty of its own. These sources do not document one. The seller-fulfilled marketplace gets a cap you can ship and confirm. FBA gets only units that have left the shared pile. If you cannot afford to cancel the shop, do not dump the residual there either. Hold it, and keep every marketplace cap inside what you will fulfil.

Reserve bundle components before a channel sees a quantity. A set that needs three of a component has promised those three for as long as the set stays on sale, even with no orders today. Subtract them in the count, or take the set down. Offering the set and the single item against the same units is an oversell you have already published.

Check the seller-fulfilled window before you raise the cap. The staff reply measures the rate over a given 7-day period and does not say that period rolls, so waiting for cancellations to leave the window is your assumption, not a rule in the reply. When seller-cancelled orders are already 2.5% of total orders, take another order on that channel only if you can ship it and confirm shipment. If you cannot, pause the offer. One more seller cancellation, including a cancellation Amazon creates for an unconfirmed shipment, moves the rate above the line in the staff reply. A paused offer takes no new orders, so later fulfilled orders bring the rate under 2.5% only if other offers on the same metric stay live and you fulfil them. Keep room for one miss before you raise the cap. A person makes that pause. Nothing in these sources describes it as an automatic switch in Amazon or in a sync tool.