Make one system the master record of stock (usually Shopify), connect Amazon and eBay to it with a sync app such as Shopify Marketplace Connect or a multichannel inventory tool, link every listing to a single SKU, and hold back a small buffer on marketplace listings so a sale on one channel reduces the others before anyone can oversell.
Three things change the setup: stock Amazon holds for you, bundles and multipacks, and sync delays at busy times. Getting it wrong costs more than a refund. eBay counts a cancellation because an item was out of stock as a transaction defect, and Amazon expects the share of orders you cancel before dispatch to stay under 2.5%.
An independent toy shop selling in four places
To keep this concrete, follow an illustrative independent toy shop. It has a physical shop using Shopify's point of sale, a Shopify online store, an Amazon account where it ships most orders itself and sends a few fast lines to Amazon's warehouses, and an eBay shop for discontinued and end-of-line toys. It carries about 1,100 SKUs (stock-keeping units, the code for each distinct product and variant).
Before automation, the owner updated Amazon and eBay quantities by hand each evening from a Shopify export. That took 40 minutes on a normal day and was hopeless in December, when the same wooden advent calendar could sell in the shop, online and on Amazon within the same hour. Last December the shop oversold 23 orders, most on the marketplaces, and spent the week before Christmas writing apology emails.
Choose the master record first
Every setup that works has one system that holds the true stock count, and every other channel reads from it. The choice depends mostly on order volume and how many places stock physically sits.
| Setup | Master record | Suits | Cost (list price) |
|---|---|---|---|
| Shopify plus Marketplace Connect | Shopify | One or two stock locations, Shopify already the main system, a few hundred marketplace orders a month | First 50 synced orders a month free, then 1% per synced order, capped at $99 a month |
| Multichannel inventory platform | The platform (Shopify becomes one channel) | Several sales channels beyond these three, shipping labels and picking from one screen | Veeqo's Inventory plan from $19 a month, priced by order volume; Linnworks is quote-based and aimed at higher volumes |
| Inventory or ERP system | That system | Several warehouses, wholesale accounts, manufacturing or kitting | Usually quote-based, plus setup |
For the toy shop, Shopify Marketplace Connect was the natural choice: the shop and online store already share Shopify stock, and the app is Shopify's own connector for Amazon and eBay. If you are weighing platforms more broadly, Veeqo's pricing page shows that its free plan covers shipping, while automated inventory sync across channels needs the paid Inventory plan.
A quick sum shows how the Marketplace Connect fee behaves. In October the toy shop had about 420 synced marketplace orders at an average of $26. The first 50 are free, so the fee is 370 x $26 x 1%, about $96. In December, with around 1,150 synced orders, 1% would be about $286, but the cap holds it at $99. At that volume the fee is effectively flat, which makes it cheap exactly when overselling would hurt most. Check in the app how orders fulfilled from Amazon's warehouses are counted for your account.
Clean the SKU map before switching sync on
A sync tool matches listings by SKU. If a listing on eBay has no SKU, or the wrong one, the tool either ignores it (so it keeps selling stock you do not have) or links it to the wrong product (so it shows a quantity from a different toy). This clean-up is the step that decides whether the whole project works, and it is where AI saves the most time.
Export three files: Shopify products with SKU, title, variant and barcode; Amazon listings with seller SKU, ASIN and title; eBay listings with item ID, custom label (eBay's SKU field), title and variation. Then give an AI assistant a strict matching job:
Match every Amazon and eBay listing below to exactly one Shopify SKU.
Match on barcode first. If there is no barcode, match on title AND variant (size,
colour, piece count, age range).
Return CSV: channel, listing_id, current_sku, matched_shopify_sku, match_basis,
confidence (high/medium/low), note
Rules:
- Never force a match. If two Shopify products are plausible, mark low and name both.
- Flag any listing whose title contains a number (pieces, pack size, age) that differs
from the Shopify title.
- Flag listings with no plausible match as UNMATCHED.
An illustrative slice of the output:
amazon,B0XXXXXX01,TRN50,TRN-050,barcode,high,
ebay,1450000001,,TRN-030,title,low,"eBay title says 50 pieces; TRN-030 is the 30-piece set. TRN-050 also plausible."
ebay,1450000002,DOLL-RED,DOLL-HOUSE-RED,title,medium,"Custom label differs from Shopify SKU"
amazon,B0XXXXXX07,PB-3PK,,none,low,"UNMATCHED: 3-pack of party bag fillers; no pack SKU in Shopify"
The second line is the classic mistake a person makes when matching 1,100 products by eye: two train sets with near-identical titles. The AI caught it because the prompt told it to compare numbers in titles. The last line is a bundle problem, covered below. For the toy shop, the AI pass took 30 minutes to run in batches and the owner spent about three hours checking the medium and low matches, against an estimated two days by hand. Then she fixed the SKUs in the marketplace listings themselves so every channel used the same code.
While you are in the listings, check that the titles fit each marketplace's limits. Amazon capped product titles at 75 characters in most categories from 27 July 2026, and eBay allows 80. How small Amazon sellers use AI for listings and writing eBay listings with AI cover rewriting them without losing the key details.
Setting up Marketplace Connect, step by step
- Install and connect. Add the app from the Shopify App Store and connect your Amazon and eBay seller accounts. Allow an hour, mostly for account permissions.
- Link existing listings. Link your current marketplace listings to Shopify products by SKU rather than creating new listings, so you keep their reviews and sales history. The clean SKU map makes this mostly automatic.
- Choose the stock location. If the shop and a back room are separate locations in Shopify, decide which ones feed the marketplaces. The toy shop let both feed online channels but held a separate allocation for shop-only display stock.
- Set inventory rules. For each marketplace you can set a fixed quantity, a buffer or a maximum quantity, individually or for groups of products. Shopify's help pages give the example that a buffer of 2 on 5 units in stock shows 3 on the marketplace.
- Decide how Amazon's warehouse stock is handled. Marketplace Connect's help pages note that orders fulfilled by Amazon on synced products do not change your Shopify counts; only orders you ship yourself do. Treat Amazon-held units as a separate pool.
- Test on a handful of products, then switch on the rest. The test plan is further down.
Buffers and caps that stop overselling
Sync is never instant, and at peak times updates queue. A buffer covers the gap: it is the number of units hidden from a marketplace so that a sale elsewhere during the delay does not cause an oversell. A simple rule is to set the buffer to the number of units that could sell on your other channels during your worst observed sync delay, plus one.
| Product type | Rule on Amazon and eBay | Why |
|---|---|---|
| Fast sellers (advent calendars, top 30 SKUs in December) | Buffer 2; maximum 10 shown | Several can sell in the shop within minutes; a cap limits damage if sync stalls |
| Normal range | Buffer 1 | Covers the occasional simultaneous sale |
| Last one or two units | Shopify and shop only | A single unit shown on four channels is the likeliest oversell |
| End-of-line toys | eBay only, fixed quantity | Keeps clearance separate from full-price channels |
| Amazon warehouse lines | Managed in Amazon; not synced from Shopify count | Separate physical stock |
You can ask AI to propose buffers from sales data: upload hourly sales by SKU and channel for last December and ask which products sold more than two units within any 15-minute window. That list is your "fast seller" group. Revisit it each season; toys go from quiet to sold out in a week when something catches on.
Bundles, multipacks and variants cause most oversells
The unmatched "3-pack of party bag fillers" is a typical problem. On Amazon it is one listing; in the stockroom it is three single packs. If the 3-pack has no SKU of its own, the sync cannot reduce the singles when it sells, and if it has its own SKU with its own count, the singles and the pack can both claim the same physical units.
- Genuine bundles (made up from singles when ordered): use a bundles app or an inventory tool that reduces the component SKUs when the bundle sells, and test it.
- Pre-packed multipacks (made up in advance and shelved separately): give them their own SKU and their own stock count, and move stock between singles and packs with a recorded adjustment.
- Variants (colours, sizes): each variant needs its own SKU on every channel. One eBay listing for "doll's house, red or pink" mapped to a single SKU will sell pink doll's houses you do not have.
A sports equipment shop selling across the same channels hits the variant problem in a different form: shoe and clothing sizes are labelled differently on each marketplace, so the same pair of trainers can appear as three different-looking variants. Map by SKU, never by the size label, and keep one size convention in Shopify.
A daily check that catches mismatches early
Even a good sync drifts: a listing gets unlinked after an edit, a marketplace rejects an update, someone changes a quantity by hand. A five-minute daily check catches it. Export available quantity by SKU from Shopify and from each marketplace (or from the app's report), put them side by side in a spreadsheet, and flag any SKU where a marketplace shows more than Shopify's count minus the buffer. The comparison itself is a spreadsheet formula; AI is useful for the next part, explaining a long list of flags and sync messages in plain language.
Paste the flagged rows and the app's recent sync errors into a chat and ask: "Group these stock mismatches by likely cause and list the fix for each group." Illustrative output:
Group 1 - listings not linked (6 SKUs): these eBay items have no SKU in the custom
label since yesterday's title edits. Re-link them to their Shopify products.
Group 2 - bundle components (3 SKUs): Amazon shows 3-packs available while single
packs are at zero in Shopify. Check the bundle rule for PB-3PK.
Group 3 - update rejected (2 SKUs): the marketplace rejected the quantity update
because the listing was inactive. Reactivate or end the listing.
Shopify's own assistant can help with the other side of the problem, running out. Sidekick can answer questions such as which products are about to run out and draft purchase orders, as covered in using Shopify Sidekick to run your store, and setting reorder points with AI explains how to decide when to reorder.
Test the sync before the busy season
Run this test in a quiet week, on five products you control, before switching sync on for everything and again before each peak:
- Sell one unit in the shop. Time how long Amazon and eBay take to show the new quantity. That is your sync delay.
- Sell one unit on eBay. Check Shopify and Amazon.
- Refund and restock an online order. Check the unit comes back on every channel.
- Cancel an unfulfilled marketplace order. Check the stock is released.
- Sell a bundle. Check every component reduces.
- Take a product to zero. Check each marketplace stops selling it rather than showing a stale quantity.
- Repeat the first step on a busy Saturday. Delays at peak are what your buffers must cover.
Write down the delays you measure. The toy shop found its quiet-day delay was a couple of minutes but a Saturday update once took much longer, which is why its fast sellers carry a buffer of 2 rather than 1.
Signs the sync has broken
Sync failures are often silent. Watch for these, and set alerts where the tool allows:
- Negative stock in Shopify, which usually means two channels sold the same unit.
- A marketplace showing stock for a product that is zero in Shopify.
- A sudden drop in sales on one channel, which can mean listings were deactivated.
- Error or warning counts rising in the app's dashboard.
- A customer message asking where an order is for a product you do not have.
The general method for monitoring automations, including who gets the alert, is in stopping automations breaking silently. Then there is the physical side: if the shelves do not match Shopify, no sync can help. A stocktake before switching sync on, and a quick count of the fast sellers each week in November and December, keeps the master record honest; running a stocktake faster with a phone scanner shows a quick way to do it.
Charity shop one-offs and wholesale orders
A charity shop listing donated items on eBay and its own Shopify store has the opposite problem to the toy shop: almost every item is a quantity of one. Buffers make no sense, because a buffer of one hides the only unit. The safe options are to list each one-off on one channel only, or to accept both channels and make sure the sync ends the other listing the moment one sells. For one-offs, a test of that exact case (sell on eBay, check the Shopify listing disappears) matters more than any other step.
A shop with a wholesale side has a third pattern: large orders that take a big chunk of stock at once. Reserve wholesale stock in a separate location in Shopify so a trade order does not suddenly zero the marketplaces mid-afternoon.
What the toy shop's setup cost, and what changed
The toy shop's costs were the Marketplace Connect fee (about $96 in October, capped at $99 in December), about five hours of the owner's time on the SKU clean-up and setup, and two hours of testing. The evening update routine disappeared. In the following December it oversold 4 orders, all on bundle SKUs set up before the bundle rule was fixed, against 23 the year before. Those figures are illustrative, but the pattern is typical: once sync is on, what remains is usually bundles and physical miscounts, which is where to look first.
The penalties are why this matters beyond saved time. Under eBay's seller standards, a seller cancelling because an item was out of stock counts as a transaction defect, and more than 2% of transactions with defects puts a seller below standard. Amazon's account health page tracks the rate of orders sellers cancel before dispatch against a target of under 2.5%. A shop that oversells in a busy week can breach both at the time of year it can least afford to.
Stock sync questions from multichannel sellers
How quickly does stock update between channels?
It depends on the tool and the marketplace, and it slows at busy times. Treat sync as near-real-time rather than instant and set buffers accordingly. The only reliable way to know your own lag is to test it: sell one unit on one channel and time how long each other channel takes to show the new quantity, at a quiet time and again on a busy day.
Should the shop till and online stock share one count?
Usually, yes, if the till runs on the same system as your online store, because every in-shop sale then reduces what the marketplaces show. If your till is separate, either connect it to the master record or keep a separate stock allocation for online sales. Selling the last unit in the shop while Amazon still shows it is one of the commonest oversell causes.
Can I let AI decide my buffer quantities?
AI can suggest buffers from your sales history, for example by looking at how many units of each product sell per hour on each channel at peak. Treat its suggestions as a starting point and review them for your fastest sellers. The final rule should be one you understand, because you will need to change it quickly when a product suddenly takes off.
Further reads
- How to Find Dead Stock and Slow Movers With AI — Clear the slow sellers clogging your listings on every channel.
- How to Automate Returns and Refunds With Clear AI Rules — Make returns go back into stock on every channel correctly.
- Can AI Read Delivery Notes and Update Stock Automatically? — Get deliveries into your master stock record without typing.
- How Much Does It Cost to Add AI to a Shopify Store? — What adding AI features to a Shopify store really costs.
- Best AI Inventory Tools for Small Retailers Compared — Compare inventory tools if you outgrow the basic sync.
- AI Inventory Forecasting for Small Businesses: How It Works — Forecast demand so the synced stock is there to sell.
- How Small Shops Can Use AI to Forecast Stock and Reorder — Clean your till export, work out one reorder point by hand, let AI run it across every line, and turn it into a 20-minute weekly ordering routine.
- How Resellers Use AI to List Second-Hand Items Faster — A second-hand listing workflow that uses marketplace AI for the drafting and keeps measurements, flaws and photos honest.
- Why AI Stock Forecasts Go Wrong: 8 Mistakes Small Businesses Make — The eight ways small businesses feed AI stock forecasts bad history or misread the output, each with a worked example and a fix you can apply this week.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Shopify App Store, Marketplace Connect listing and Shopify help pages on Marketplace Connect listings, inventory rules and fulfilment; Veeqo pricing page; eBay seller standards policy; Amazon Seller Central guidance on pre-fulfilment cancel rate.