Why do I keep selling stock that is already gone, and how do I stop it?

What overselling is, and why it hurts

Overselling means a customer pays for something you no longer have. Each case costs you a refund or a delay, an apology, and often a bad review. On Amazon it can also hurt your account health. Amazon tracks a seller-fulfilled metric called the pre-fulfillment cancellation rate, which counts orders you cancel yourself over a 7-day period. Amazon's stated target is under 2.5%, and going above it can put an at-risk banner on your account health page, which Amazon says does not by itself mean action is required. Orders cancelled by the customer are not counted. Only the ones you cancel are.

The small-store maths is harsh. At 40 orders in a week, one cancelled order is 2.5%, right at the target. Overselling does not need to happen often to matter.

The real causes

1. Sync lag

You sell the last unit on Shopify (a platform for running an online shop). Your tool then tells Amazon, but only every 15 minutes, or only after a nightly run. In that gap the same unit sells again. The busier the product, the more this hurts.

2. Stock counted in more than one place

One channel holds 10 units. Another holds a separate 10. Your warehouse has 12. Nobody reconciles the three numbers. Selling "the same item" on three channels without one shared count is the most common cause I see.

3. Bundles and kits

A gift set is made of a candle, a matchbox and a card. If someone buys the set, the candle count must drop too. Many shops forget this link, so the set shows as in stock while its parts are gone, or the reverse. I cover this in more detail in my guide to Shopify bundle inventory.

4. Manual spreadsheets

A shared spreadsheet works until two people are in a hurry. One forgets to deduct a return. Another types a count into the wrong row. A sheet is a record, not a system, and it never updates a channel by itself. If you are living with one, my guide to a multichannel inventory spreadsheet shows how to make it safer.

5. Settings that allow it

By default, Shopify stops customers buying an item once tracked inventory reaches zero. But each product or variant has a "Continue selling when out of stock" option, which Shopify offers for pre-orders and for selling before you reorder. If it is ticked on a product that is not a pre-order, the shop will take orders with nothing behind them. The option only works when "Inventory tracked" is on, so an untracked product never stops selling. Orders rung up on Shopify POS (its in-person till) are not blocked at zero either: staff get a warning but can still sell. With more than one location, an item can show as out of stock online while another location still holds units, if that location does not fulfill online orders. Check all of this before you blame a sync tool.

If you ship from more than one place, look at order routing too. Shopify applies routing rules in order to decide which location fills each order, and by default it prefers a location that can fill the whole order and the one closest to the customer. It needs two or more active locations. Those rules decide which location's count drops, so a surprise "out of stock" often starts there. Shopify Flow, its free automation app, also has triggers named "Product variant out of stock" and "Product variant inventory quantity changed", which you can use to message yourself the moment a count hits zero.

6. Returns, damage and FBA gaps

A returned item goes back on the shelf in your head but not in your system. A damaged unit is still counted. If you use Amazon FBA (Fulfilled by Amazon, where Amazon stores and ships your goods), stock in transit to a warehouse can be double-counted. My guide to Amazon FBA inventory problems goes through those gaps.

The fixes, in the order I would do them

1. Pick one stock record. One number per product that every channel reads from. This is the single most useful change. It can be an inventory tool, or a small system built for you. See my guide to an inventory management system for how these work.
2. Sync both ways, as often as the tool allows. A stock change on one channel should reach the others in minutes. My stock sync for Shopify guide covers the options.
3. Hold a buffer on fast sellers. A buffer is a few units you hide from sale on each channel. If you have 20 units and a buffer of 3, channels show 17. Use a bigger buffer for items that sell fast and a smaller one for slow items.
4. Break bundles into parts. Make the system deduct each component when a bundle sells.
5. Set low-stock alerts. An alert at a set level, such as 10 units or 7 days of sales, gives you time to reorder or pause a listing. Without alerts you learn about the problem from a customer.
6. Do a weekly count on your top sellers. Ten minutes a week catches drift before it grows. Count the 10 products that make most of your sales.
7. Write the returns rule down. Say who checks a returned item and when it goes back on sale.

A worked example (made-up figures)

A skincare brand sells on Shopify, Amazon (seller-fulfilled) and Etsy (a marketplace for handmade and vintage goods). A serum has 12 units on the shelf. Sync runs every 15 minutes. On a busy Saturday it sells 5 on Shopify, 4 on Amazon and 4 on Etsy within that window. That is 13 units against 12 in stock, so one order is oversold.

With a buffer of 3, each channel sees 9 units. The first nine sales across all channels succeed, the next sync hides the listings, and the three held-back units cover the lag. Sync can also be shortened, but the buffer is the cheap, quick safety net. The brand then sets a low-stock alert at 6 units so the reorder starts before the Saturday rush.

When it signals a bigger process problem

If you oversell even with sync and buffers, the cause is usually upstream. Watch for these signs:

• You do not know your real stock until you count.
• Different staff update the numbers in different tools.
• Your purchasing runs on feel, not on sales rates.
• You add a channel without adding a stock rule.

Those mean the process needs fixing, not just the settings. My guide to business process automation shows how to map and tidy that flow. If AI interests you, I wrote a separate piece on whether AI can help stop overselling.

How I'd approach this

I start with a list of the last 20 oversold orders. For each one, I mark which of the six causes it was. Nearly always, two causes explain most of them. Then I fix those two first. For a shop on two or three channels, my usual build is a single stock record, a buffer rule per product, and a low-stock alert that lands in email or a chat message. That is the core layer I build at Blur Studio, and you can see how it fits your shop on a quick call with me. If a ready-made tool covers your case, you should use that and skip the build.