What Makes a Winning Product? The 4 Filters Every Seller Should Use
What Makes a Winning Product? The 4 Filters Every Seller Should Use
We quote several hundred product briefs a year out of Yiwu, and we can usually tell within ninety seconds whether the client is going to make money on it. Not because we know their market better than they do - we do not - but because a winning product has to satisfy four constraints simultaneously, and most briefs fail at least one of them before a single sample is made. The four filters are margin, durable demand, shipping friendliness and video demonstrability, and they apply whether you sell on a marketplace, your own store, wholesale into retail, or all three.
Why 'Winning Product' Lists Keep Failing You
The internet is full of lists of winning products. They fail for a structural reason: a product is not intrinsically good or bad, it is only good or bad relative to a cost stack, a channel and a customer. A collapsible silicone dog bowl is a superb product for a lightweight direct-to-consumer brand and a terrible one for a heavy-goods wholesaler. The list cannot know which you are. A filter framework can, because it forces you to test the product against your own constraints rather than against somebody else's screenshot.
Filters are for elimination, not selection
Run them in order and use them to say no. Ninety percent of candidates should die at filter one or three. The goal is not to find a product you like - it is to remove every product that cannot survive your cost structure.
The Four Filters, In Order of Elimination
Order matters. Run margin first because it is the cheapest to check and kills the most candidates. Run demand second because it needs a five-year data pull. Run shipping third because it requires a physical spec from a supplier. Run the video test last because it is subjective and only worth your time on a product that has already survived the other three.
| Filter | The test | Kill threshold |
|---|---|---|
| 1. Margin | Landed cost supports a 3-5x retail multiple | Below 3x on realistic landed cost |
| 2. Durable demand | Five-year trend shape is rising or stable | Single decayed spike, or novelty-dependent |
| 3. Shipping friendliness | Under 2 lb, low cube, non-fragile, no battery | Fragile, oversized, or restricted |
| 4. Video demonstrability | Benefit visible in a 15-second clip | Requires explanation to be understood |
Filter 1 - Margin: The 3-5x Landed Cost Multiple
The multiple is the fastest sanity check in sourcing. Take the realistic landed cost - factory price plus per-unit freight plus the duty stack, not the FOB price alone - and ask whether the market will bear three to five times that number. Below 3x you have no room for advertising, returns or a freight shock. Above 5x you should ask why the market allows it, because the answer is usually a brand, a patent, or a customer you cannot reach cheaply.
Concretely: a product landing at $6.40 needs to retail credibly between $19 and $32. If comparable items on your channel transact at $14, the product is dead at filter one no matter how much you like it. The multiple is not a pricing strategy, it is a feasibility test.
Use landed cost, never FOB
The single most common error we see is applying the multiple to the factory quote. A $4.20 FOB item is not a $12.60 product at 3x; it lands nearer $6.40 once freight and duty are counted, so 3x is $19.20. Sellers who multiply FOB set their retail price 35-40% too low and then wonder why advertising is unprofitable.
How to Test the Multiple Before You Order
You can validate the multiple in under an hour without buying anything.
Get a real FOB quote with weight and carton data
Ask for FOB price, net weight, gross weight, carton dimensions and units per carton. A supplier who cannot supply carton data has not exported this item.
Estimate per-unit freight
Divide carton cube into container or courier rates. For small parcels, price both express air and consolidated sea - the gap is often 3-4x per unit.
Add the duty stack
Confirm the HTS code with a broker and add every applicable duty line, plus the small processing fees.
Find the transacted price ceiling
Use sold-listing data rather than asking prices to establish what the market actually pays, not what sellers hope for.
Divide
If the ceiling divided by landed cost is under 3, stop. You have spent an hour and saved a container.
Filter 2 - Real Demand, Not a Fad
Fads are seductive because the early numbers look extraordinary. The problem is manufacturing lead time. From a signed purchase order, a typical Yiwu run takes 20-35 days to produce and 25-40 days on the water to the US or EU. That is roughly two to three months between deciding and selling. A fad with a six-week lifespan will be over before your container is unloaded, and you will own the inventory.
The three demand shapes
- Rising staircase - steady growth over years, each year higher than the last. This is a real trend and you can order into it confidently.
- Repeating wave - a clean annual seasonal cycle. Perfectly viable, but your order calendar has to lead the wave by three months, not follow it.
- Single spike with decay - vertical rise, then a long fade. This is a fad. If you did not order before the spike, you are too late, and the people selling it now are liquidating.
The distinction between durable demand and hype is worth studying in its own right, because the two look identical for the first four weeks. The reliable tell is whether the product solves a problem that existed before the product did. A silicone lid that stops a pan boiling over solves a problem from 1850. A novelty that trends because of a dance solves nothing.
If the problem existed before the product did, the demand will outlive the trend that revealed it.
Filter 3 - Shipping Friendliness: Under 2 lb and Non-Fragile
This is the filter that separates people who have imported from people who have only researched. Physical properties dominate the economics of small-parcel products, and four of them matter.
- Weight. Aim for under 2 lb (907g) shipped, ideally under 1 lb. Fulfilment fees and courier rates step in bands, and crossing a band costs real money on every unit forever.
- Cube. A 40HQ container gives roughly 67-68 usable CBM. A product that nests or collapses can double the units per container, halving freight per unit without changing anything else.
- Fragility. Glass and ceramic run breakage of several percent through a multi-touch supply chain, plus the customer-service cost of every broken arrival. Polymer and metal equivalents run a fraction of a percent.
- Restrictions. Lithium batteries, aerosols, magnets, liquids and anything with a plug all add certification, restricted shipping modes and rejected air freight. Every one of them is a compliance project, not a product.
The weight-band cliff nobody warns you about
We have seen a product go from viable to unviable over 40 grams of packaging. The unit weighed 890g and the retail box brought it to 947g, which crossed a fulfilment weight band and added roughly $0.90 per unit. On a 15,000-unit year that is $13,500 - lost to cardboard. This is why we ask for gross weight, not net weight, and why our first cost-reduction move on almost any product is packaging. RND SOURCING has recovered more margin for clients with a box redesign than with any price negotiation.
Ask for gross weight in the first email
Net weight is the product. Gross weight is what you pay to move. Request net weight, gross weight, carton dimensions and units per carton on every quote request - we put all four on our standard inquiry sheet for exactly this reason.
Filter 4 - Does It Demonstrate in 15 Seconds?
Whatever channel you sell on, the buying decision is now visual and fast. A short-video-friendly product has a visible before-and-after: something transforms, collapses, cleans, seals, organises or fixes in a way a viewer understands with the sound off. Products that require explanation - a slightly better cable, a marginally softer fabric - can be excellent businesses, but they need a brand, an audience or a retail shelf, and they will not carry themselves on paid social.
The honest test: describe the product's benefit to somebody without using an adjective. If you can only say 'it is nicer', it will not demonstrate. If you can say 'it folds flat to the size of a phone', it will.
The Four-Dimension Scorecard
Score each filter 0-5 for a maximum of 20. Our rule with clients: proceed at 14 or above with no single dimension below 2. A product scoring 18 on three dimensions and 1 on shipping is not an 'almost' - it is a no, because the failing dimension is structural and will not improve.
| Dimension | Score 1-2 | Score 3 | Score 4-5 |
|---|---|---|---|
| Margin multiple | Under 3x landed | 3x to 3.5x | Above 4x with headroom |
| Demand durability | Decaying spike | Flat but stable | Multi-year rising or clean seasonal |
| Shipping friendliness | Fragile, over 3 lb, or restricted | 1-2 lb, sturdy | Under 1 lb, nests or collapses, unbreakable |
| Video demonstrability | Needs explanation | Visually pleasant | Obvious transformation in 15 seconds |
Worked Example: Scoring Three Real Candidates
Three briefs we received in the same month, scored with the framework. All three were products the client was excited about.
| Candidate | Margin | Demand | Shipping | Video | Total |
|---|---|---|---|---|---|
| Ceramic pour-over kettle, 1.4 kg | 4 | 5 | 1 | 3 | 13 - declined |
| Collapsible silicone pet bowl, 180g | 4 | 4 | 5 | 4 | 17 - proceeded |
| Rechargeable LED desk lamp, 1.1 kg | 3 | 4 | 2 | 4 | 13 - declined |
The kettle had genuinely excellent demand and a healthy multiple, and we still advised against it: 1.4 kg of ceramic through a multi-touch chain produced a breakage and replacement cost that ate the margin the multiple promised. The lamp failed on a lithium cell that restricted air freight and required certification the client had not budgeted for. The pet bowl - the least exciting of the three - collapsed to 12mm, weighed 180g, was effectively unbreakable, and demonstrated itself in one motion. It is still selling three years later.
What to Do When a Product Fails One Filter
A failed filter is not always the end. Two of the four are fixable at the factory, and two are not.
- Margin failure - often fixable. Re-spec materials, cut cube, consolidate freight, verify the duty code. We routinely move a product from 2.7x to 3.6x without touching the supplier's unit price.
- Shipping failure - sometimes fixable. A fragile ceramic item can become a coated stoneware or polymer item; a rigid item can be redesigned to nest. A lithium battery usually cannot be removed without becoming a different product.
- Demand failure - not fixable. You cannot manufacture durable demand for a fad. Walk away.
- Video failure - not fixable by sourcing. It is a marketing and channel problem, solvable with a brand or a retail buyer, but not with a better factory.
This is precisely the conversation worth having before you tool up rather than after. When a client sends us a candidate that scores 12, our job is to tell them which of the four dimensions can be engineered and which one is fatal. Two decades on the Yiwu floor mostly teaches you which failures are worth fighting.
Conclusion
A winning product is not a clever idea; it is a product that simultaneously clears a 3-5x landed multiple, sits on durable rather than novelty demand, is light and unbreakable enough to move cheaply, and explains itself on screen in fifteen seconds. Score every candidate out of 20, demand at least 14, and refuse anything with a structural failure in a single dimension. Send your top three candidates and RND SOURCING will supply the real weights, carton data and landed costs you need to score them honestly.
What makes a winning product to sell?
Four things at once: a landed cost that supports a 3-5x retail multiple, demand that is durable rather than a passing fad, physical properties that ship cheaply (under 2 lb, low cube, non-fragile, unrestricted), and a benefit that is visible in a 15-second video.
What profit multiple should I aim for on a product?
Three to five times realistic landed cost - not FOB price. Below 3x there is no room for advertising, returns or freight volatility. Above 5x, check whether a brand, patent or hard-to-reach customer is the real reason the price holds.
Why does product weight matter so much?
Fulfilment fees and courier rates step in weight bands, so crossing a band adds cost to every unit permanently. We have seen 40g of packaging add $0.90 per unit. Always request gross weight, not net weight, when quoting.
How do I tell a real trend from a fad?
Pull five years of trend data and read the shape. A rising staircase is real, a repeating annual wave is seasonal, a single spike with a long decay is a fad. Also ask whether the problem existed before the product did - if it did, demand outlives the trend.
Is this framework only for Amazon FBA sellers?
No. The four filters apply to any physical-product channel: your own store, marketplaces, wholesale into retail, or social commerce. Only the specific fee numbers change; margin, demand durability, shipping economics and demonstrability constrain every channel.
Use the filters to eliminate, not to fall in love. Score out of 20, insist on 14, and treat a single structural failure as fatal. Talk to RND Sourcing and we will supply the weights, carton cube and landed costs that make an honest score possible before you commit a mould.
