Finding a product that appears to sell well is only the start of sourcing. The seller must obtain genuine, eligible units at a cost that leaves a margin after marketplace and fulfillment charges, and the supplier must be able to repeat the result. A popular search term does not validate a particular factory, listing, or purchase order. Validation turns a promising idea into a small, documented test with a clear stop rule.
Research the buyer need, verify the product and supplier, and buy the smallest quantity that can answer the important questions. This third article in the Amazon Selling Guide builds on the selling-model choice and eligibility gate. The later guides cover fees and cash flow, listings and inventory controls, and performance review. The examples describe a US seller; terms, documents, and customer demand must be checked for the exact product and marketplace.
Start with a customer problem and a product specification
Write what the buyer is trying to accomplish, the product features needed, the alternatives available, and where those alternatives disappoint. “High demand for kitchen goods” is too broad. “Small apartments need a durable utensil holder that fits a narrow counter and cleans easily” is testable. The product specification should include dimensions, material, capacity, color, package contents, safety or performance requirements, and price range. Without that specification, supplier quotes and catalog comparisons can describe different products while appearing interchangeable.
If reselling an existing item, identify the exact brand, model, variation, pack count, and condition. A cheaper supplier’s unit is not a substitute if it has different labeling, region, accessories, or warranty terms. If creating an original product, write the quality acceptance criteria before requesting samples. A sample that looks attractive in a photograph may fail a size, durability, or labeling requirement. The product restriction guide should already have shown whether the category, brand, specific ASIN, and intended fulfillment method are eligible.
Use customer reviews as qualitative evidence, not as instructions to copy another seller’s design or protected content. Look for recurring complaints that a lawful product improvement could address. Separate defects from buyer misuse and shipping damage. A review set can be biased toward unusually satisfied or dissatisfied people. Ask whether your proposed improvement is measurable and whether buyers will notice it enough to choose your offer at a viable delivered price.
Read marketplace research with its limits
Amazon’s Product Opportunity Explorer groups search terms and products into niches and can show customer demand and unmet-need signals. It is a useful discovery tool, but a niche-level signal is not a purchase order for your particular product. Check relevant listings, variations, prices, reviews, seasonality, and competing substitutes. A large market can be difficult to enter when established offers already satisfy the buyer; a small market can be worthwhile if your costs and customer fit are strong.
Record what the data actually measures. Search volume is interest, not your expected sales. A competitor’s estimated or historical sales do not transfer to a new listing with no reviews and different delivery terms. A bestseller rank is relative and can move quickly. Sponsored placements can make a page appear more attractive than organic demand would justify. The goal is to form a conservative sales range and an experiment, not to convert one dashboard number into a guaranteed forecast.
Compare the full buyer offer. A $30 competitor may include a two-pack, next-day delivery, a recognized brand, or a generous warranty. Your $27 single unit may not be cheaper in practical terms. Check the date and whether a price is a temporary promotion. Look at low, middle, and high outcomes rather than selecting the most favorable example. For a new product, consider the cost of photography, content, samples, launch inventory, and customer acquisition. The fee article will quantify those costs before a scale decision.
Interview or observe real users when possible. A short conversation about how a buyer currently solves the problem can reveal a wrong assumption in a keyword analysis. Ask about purchase criteria and failures, not “Would you buy my product?” as a leading yes/no pitch. For a product you already sell elsewhere, actual return reasons and repeat purchases may be stronger than a generic marketplace trend. Document the source and date of each insight so you can revisit it after a test.
Verify the supplier, not only the unit price
For wholesale, check that the supplier is a real business with a traceable address, contact channel, and relationship to the brand or manufacturer. Amazon’s wholesaler guidance recommends validating suppliers and maintaining purchasing invoices. Ask whether the supplier is authorized to sell the product to you for Amazon resale, what territory and channel limits apply, and whether documentation can be provided for an approval or authenticity question. A middleman’s assertion that goods are “100% original” is weaker than a verifiable chain of supply.
Request a written quote that states exact SKU, units, unit price, minimum order, payment terms, packaging, shipping terms, delivery date, defect allowance, returns or credits, and who bears loss in transit. A low unit price can be undone by freight, duties, inspection, labeling, prep, or a high unusable-unit rate. If the supplier changes a product component or package, require notice. The product you approved and photographed must match the product you eventually receive.
For private-label manufacturing, assess the factory’s ability to meet a specification repeatedly. Ask how it controls materials, tests batches, handles failures, and documents compliance. Request samples from the production process you intend to use, not only a polished marketing sample. If claims such as “food safe,” “child safe,” or “medical grade” matter, identify the applicable legal standard and credible evidence before using them. A supplier certificate without scope, date, product match, or issuer verification may not support a listing claim.
Be cautious with directories, marketplaces, and unsolicited pitches. A website can display a factory photo it does not own. Verify business identity through independent sources and communication methods. Do not send a full production payment to a new supplier solely because a chat agent offers a deadline discount. If a supplier refuses a small sample order, basic documentation, or a clear contract, that is a sourcing risk. A source should be evaluated by what it can deliver and substantiate, not its most optimistic price sheet.
Calculate landed cost and a maximum order
Landed unit cost includes the purchase price plus freight, import duty where applicable, receiving, inspection, prep, labels, packaging, and the cost of unusable units. For a 100-unit order costing $500 plus $100 freight, $40 prep, and $60 expected damage or nonconforming units, the practical cost is not simply $5 per unit. Divide the full relevant cost by the number of saleable units. If only 90 pass inspection, the effective cost per saleable unit is substantially higher. Keep assumptions explicit; a future order may have a different yield.
Run Amazon’s Revenue Calculator for the exact product dimensions, weight, category, price, and chosen fulfillment method. Add supplier and operating costs that the calculator does not know. Test a normal price, a realistic competitive markdown, and a return or slower-sales case. A product that only works at the highest current competitor price or perfect yield is a fragile first purchase. The fees and cash-flow guide provides the fuller contribution and timing model.
Set a maximum cash commitment before negotiating quantity discounts. A lower unit price at 1,000 units may be worse than a higher price at 50 units if the larger order consumes bill money and sits in storage. Include samples, inspection, and a reserve for returns and reorders. If the minimum order exceeds your safe test budget, do not quietly expand the budget; look for a different supplier, product, or channel. A test that can fail safely is more valuable than an impressive purchase order that must succeed.
Document the supplier’s return and defect remedies. “We replace bad units on the next order” is less useful if you may not place a next order. Clarify who pays return freight, how defects are evidenced, how quickly credit is issued, and whether a specification dispute can be resolved. The cash cost of a failed batch should be modeled before payment. Supplier reliability is part of margin, not a separate soft concern.
Inspect samples and the first batch
Order samples when the product or source is new. Test the actual buyer experience: packaging, instructions, function, dimensions, smell, finish, durability, and whether the item arrives undamaged. Compare multiple samples where batch consistency matters. Photograph markings and packaging. If the sample does not meet your written specification, do not assume production will be better. Ask for a corrected sample or stop.
When inventory arrives, count it and inspect a meaningful portion according to the product’s risk. Confirm model, color, pack count, labels, expiration dates where relevant, and damage. For critical safety or function requirements, a sampling approach may be insufficient; use the appropriate inspection standard. Segregate failed units so they cannot be accidentally offered as new. Record batch or lot numbers when provided, especially for products with safety or recall implications.
If sending stock to FBA, check preparation, labeling, carton, and shipment requirements for the actual product. Do not seal a supplier’s pallet and send it to Amazon without verifying that it contains the approved SKU and quantity. A wrong count or label can create stranded inventory or customer complaints. If seller-fulfilled, assign each saleable unit a physical storage location and ensure the offer quantity matches what you can ship. The listing and inventory controls guide covers that operating handoff.
Keep the sample and first-batch results tied to supplier and purchase order. If a later batch differs, you can identify what changed. Save invoices, product specifications, test evidence, and receiving records in a way that can support an account or customer question. Amazon’s authenticity-document guidance emphasizes genuine, matching invoice information when such evidence is requested. Requirements vary by case, so read the current notice rather than following a generic document checklist blindly.
Set a small paid test and a stop rule
A useful first order is large enough to test listing, fulfillment, returns, and buyer response, but small enough that a weak result does not threaten ordinary finances. Choose an observation period and a minimum acceptable retained margin before buying. Decide what would cause you to stop: inability to list, a material product defect, too many returns, actual fees above the model, no credible buyer demand, or a supplier unable to repeat the same unit. A clear stop rule prevents “just one larger order” from becoming the response to every warning sign.
Track units through a funnel: ordered, received, passed inspection, eligible to list, live, sold, retained after return window, and still in stock. Calculate margin for the whole cohort, not only the first few successful orders. If ten units sell profitably but ninety sit in storage, the early orders do not prove the whole buy was wise. If a product sells slowly but consistently at a strong margin, a smaller replenishment may be better than a large rush order.
Reorder only after confirming supplier price and quality, current account eligibility, current fee estimate, and real sales pace. A product that worked last quarter can fail after a fee change, competitor entry, supplier substitution, or new restriction. Do not use a forecast built on launch-week promotion as a normal run rate. The performance guide helps decide whether to expand, revise, pause, or exit after the test.
If the test fails, dispose of stock responsibly and keep the learning. A small negative result may reveal a product specification that buyers do not value, a supplier with unreliable quality, or a channel mismatch. Those are distinct causes. Do not blame advertising when customers return a flawed item, and do not blame the supplier when the market price was always below the fully loaded floor. Correct the cause before another purchase.
Separate a repeatable product from a temporary opening
A clearance or seasonal gap can be profitable once without supporting a durable business. Write down why the opportunity exists and how long it may last. If one retailer is liquidating old packaging, there may be no replenishment at the same price. If a competitor is out of stock, its return can change your price and sales pace. If a search trend follows a short event, the demand may fade before your next shipment clears receiving. A supplier’s willingness to sell more units does not establish buyer demand for them.
For a repeatable product, ask whether the supplier can provide the same specification, quality, documentation, and lead time for at least one more cycle. Get a new written quote rather than assuming the pilot price will continue. Compare the reorder margin using current fees and a less favorable selling price. If the product still works, decide how much inventory covers realistic demand without tying up cash for months. A reorder point should reflect both sales pace and lead-time variation.
For a one-time opportunity, set an exit date and a conservative price path before buying. A small lot may make sense if you can list and fulfill it quickly. Do not build fixed monthly expenses around it or present it as stable income. Keep its results separate from ongoing products in the performance review. Otherwise, one lucky liquidation can make average margin look stronger than the business can repeat.
The same distinction applies to private label. A launch promotion may generate trial orders, but repeat purchases and organic buyer response are different evidence. If the product needs continuous paid ads merely to move units, include that cost in the normal margin. If customer complaints point to a design flaw, the next batch needs a corrected specification and fresh validation, not just new photos.
Common questions about product sourcing
Is Product Opportunity Explorer proof that a new product will sell?
No. It shows marketplace research signals at the niche and search level. Your particular product still needs a buyer-relevant difference, compliant listing, competitive delivered offer, sufficient margin, and a real fulfillment plan. Use the tool to form a test, not a guarantee.
Is the lowest wholesale quote the best source?
Only after freight, defects, documentation, authorization, consistency, payment terms, and returns are counted. A slightly higher-cost supplier with reliable genuine goods and usable invoices may produce a better net result and lower account risk.
Should I buy a larger quantity to unlock a discount?
Only if the savings exceed the extra holding, storage, cash, and obsolescence risk under a conservative sales forecast. Do not let a discount force a commitment beyond the safe test budget.
When is a product validated?
After it passes eligibility and quality checks, reaches buyers, produces retained sales at acceptable contribution, and can be supplied again at similar terms. A sample, a high search volume, or one early order is promising evidence but not the whole validation.
The best source is one you can verify and repeat, supplying an exact product that a buyer wants at a cost the full transaction can support. Begin with a specification and a small order, inspect what arrives, and let retained sales—not supplier promises—decide the next purchase.