A seller walks a Target clearance endcap and finds a discontinued LEGO set marked down to $24.99. The same set is selling on Amazon for $54.99 with three offers. She buys all eleven on the shelf, ships them to FBA, and clears roughly $13 a unit after fees. That, in one trip, is retail arbitrage: buying products at retail below their market price on another channel and reselling them, usually on Amazon.
The model still works in 2026, but it is less forgiving than the YouTube version. Margins are thinner, brand gating is broader, IP enforcement is more aggressive, and Amazon now reimburses lost inventory based on your sourcing cost, which means your bookkeeping is no longer optional. This guide covers how the model actually works, where the money leaks out, and the math you should run before scanning a single barcode.
How Retail Arbitrage Works, and How Online Arbitrage Differs
Retail arbitrage (RA) is in-person sourcing: you walk physical stores, scan items with the Amazon Seller app or a third-party scouting app, and buy anything where the spread between shelf price and Amazon price covers fees plus an acceptable profit. Online arbitrage (OA) is the same trade executed from a desk: you buy from retailers' websites, often stacking coupon codes, cashback portals, and credit card rewards on top of the sale price, then ship to a prep center or your garage.
Amazon arbitrage in either form is fundamentally a deal-flow business. You are not building a brand or a moat; you are getting paid for finding mispriced inventory faster than other sellers. RA tends to win on margin because clearance in a physical store is invisible to most of the internet. OA wins on volume and repeatability because a good online deal can be bought in quantity and re-bought when it recurs.
- •RA strengths: deeper discounts, less competition per deal, no shipping cost to acquire
- •RA weaknesses: time-intensive, hard to delegate, inventory limited to what is on the shelf
- •OA strengths: scalable from anywhere, stackable discounts, easier to systematize with software
- •OA weaknesses: every deal is visible to everyone running the same tools, so prices compress fast
Sourcing: Clearance Cycles and OA Tools
Profitable RA sourcing is mostly about knowing retail clearance rhythms: seasonal resets, planogram changes, and store-specific markdown schedules. Experienced sellers learn which chains mark down on which cadence and check price-tag conventions that signal a final markdown. The job is unglamorous, repeated routes through the same stores, but the spreads on end-of-life products can be large precisely because nobody else is standing in that aisle.
OA sourcing runs on software. Scanning tools crawl retailer sites and match products to Amazon listings, flagging items where the spread looks profitable; sourcing lists sell pre-vetted deals by subscription. Both work, with a caveat: any deal delivered to hundreds of subscribers will see its Amazon price sag as those sellers' inventory lands. Treat tool output as leads to verify, not buy signals. Check the price history, the number of sellers on the listing, and whether Amazon itself is on the buy box before committing cash.
Whatever the channel, the discipline is the same: set a minimum ROI floor (many arbitrage sellers use something like 30 to 50 percent on cash invested as a working rule, adjust for your own costs) and a minimum dollar profit per unit, and walk away from anything below it. Thin deals consume the same prep, shipping, and accounting effort as fat ones.
Ungating and IP Complaints: The Risks Nobody Prices In
Many categories and brands on Amazon are gated: you need approval before you can list. Approval typically requires invoices from a distributor or manufacturer, and retail receipts from a clearance run often do not qualify. That means a portion of what you find in stores simply cannot be listed under your account, and you need to check gating status in the Seller app before you buy, not after.
The bigger tail risk is intellectual-property complaints. Reselling genuine products you lawfully bought is generally protected by the first-sale doctrine in the US, but Amazon is not a courtroom. When a brand files an IP or inauthenticity complaint, Amazon usually deactivates the listing first and asks questions later, and the standard request is invoices proving supply-chain provenance. A stack of store receipts may or may not satisfy that request. Some sellers resolve complaints; others eat the inventory and, in worse cases, account-level damage. None of this is legal advice; if you face an IP complaint or have questions about resale rights in your situation, consult a qualified attorney.
The practical rules: avoid brands known to enforce aggressively against resellers, keep every receipt and order confirmation organized by purchase lot, and never let a single brand become a large share of your inventory value. An IP complaint on 5 percent of your catalog is a bad week; on 40 percent it is an existential event.
See what Amazon owes you — free
Connect your seller account and get a free reimbursement audit. No credit card, keep 100% of what you recover.
The Margin Math: A Worked Example After Amazon Fees
Here is a realistic, fully illustrative example. You find a toy on clearance for $18.00 that sells on Amazon for $49.99. Amazon's referral fee runs 8 to 15 percent of sale price depending on category; toys sit at the top of that range, so call it $7.50. FBA fulfillment for a standard-size item of this weight might run around $6.50, and inbound shipping plus prep adds roughly $1.50 per unit; both vary, so always check Amazon's current fee schedule rather than trusting a rule of thumb. Budget another $1.00 per unit as a returns and storage allowance.
That leaves $15.49 of profit on $18.00 of cash deployed: about a 31 percent net margin and an 86 percent ROI. That is a genuinely good arbitrage deal. Now notice the sensitivity: if the price drops to $42.99 while your units are in transit, which happens constantly when other sellers find the same clearance, your profit falls to roughly $9.50. At $36.99, you are near breakeven. Arbitrage selling lives and dies on this fragility, which is why per-unit dollar minimums matter more than ROI percentages alone.
Why Arbitrage Sellers Need Landed Cost Per Lot
A wholesale seller buys the same SKU at the same price every month. An arbitrage seller might buy the same ASIN at $18 in March, $24 in April from a different store, and $15 in May with a coupon stack. If you record one blended guess as your cost, every margin report you ever run is fiction, and you cannot tell which sourcing trips actually made money. The unit of truth in arbitrage is the lot: this purchase, at this price, with these acquisition costs attached.
Since March 31, 2025, this stopped being just a reporting issue. Amazon now values FBA reimbursements at the seller's manufacturing or sourcing cost, using its own estimate unless you provide your actual cost, and excluding your margin and fees. Pair that with the October 2024 change that cut the fulfillment-center claim window to 60 days, and the seller who cannot produce per-lot costs quickly gets reimbursed at Amazon's guess, on Amazon's clock. This is the specific problem BeanHawk's PO and landed-cost engine is built for: it tracks cost per lot, maintains a perpetual valuation of your FBA inventory, and its free reimbursement audit checks what Amazon owes you with no card required and 100 percent of recoveries kept.
Even if you use a spreadsheet instead, the rule stands: record date, store, receipt, quantity, unit cost, and allocated inbound cost for every buy, the day you make it. Lot-level cost data is also your invoice trail when an IP complaint or ungating request lands.
Scaling Limits: Arbitrage vs Wholesale
Arbitrage has a ceiling, and it is made of your own hours. Every dollar of profit requires a new sourcing decision; nothing replenishes itself. RA scales only as fast as you can walk stores or train and trust someone else to. OA scales further, but the deals that software can find are the deals everyone's software can find, so competition arrives within days and margins compress toward zero on popular finds. Capital also recycles slowly: cash sits in transit, in FBA receiving, and in the payout cycle before it can be redeployed.
That is why experienced sellers treat arbitrage as a training ground and a cash engine rather than a destination. It teaches you fee math, listing dynamics, and cash-flow discipline with low stakes. The common graduation path is wholesale: opening accounts with distributors or brands, buying replenishable inventory on terms, and turning sourcing from a daily hunt into a purchase order you reorder monthly. The skills transfer directly, and so does your cost-tracking habit, because wholesale margins are thinner and only survive with accurate landed costs.
- •Signs you have outgrown arbitrage: your hourly profit is flat while hours grow, your best ASINs keep getting gated, and unsold tail inventory is accumulating
- •What to carry into wholesale: your fee-math discipline, per-lot cost records, and relationships with categories you know
- •What to leave behind: chasing every spread, single-unit buys, and inventory you cannot document
The Bottom Line
Retail arbitrage in 2026 is a real but narrow business: profitable for sellers who run the fee math before buying, document every lot, respect gating and IP risk, and treat it as a stepping stone rather than a permanent model. The sellers who fail are rarely bad at finding deals. They are bad at knowing, with numbers, whether the deals they found actually made money after fees, returns, price drops, and lost inventory. Get the cost side right from your first sourcing trip and every later decision, from which stores to revisit to when to jump to wholesale, gets easier.