Choosing a model and your first product

Product research is sold as the hard, mystical part of this business. It is neither. It is a sequence of eliminations, and the skill is in throwing things away quickly. A beginner spends three weeks falling in love with one idea. Someone who has done this a while rejects forty products in an afternoon and spends the evening on the two that survived.

This lesson builds that sequence: first the model, then the category, then the physical object, then the restrictions, then the competition, and only at the end the arithmetic. Doing the arithmetic first is the most common way to waste a weekend, because a product that clears on margin can still be one you are not allowed to sell.

Step 1. Fix the model before you look at any product

Lesson 0 described four models. Choose using three constraints, in this order.

1. Cash you can lose. If that number is small, private label is off the table for now — not because it is bad, but because it commits everything before you learn anything.
2. Hours per week, and when they fall. Arbitrage is paid by the hour in disguise: stop hunting and the income stops. Wholesale is mostly correspondence, which fits around a job. Private label is bursts of intense work separated by weeks of waiting.
3. What you want in two years. Arbitrage does not compound; the deal you found is gone next week. Wholesale compounds through supplier relationships. Private label compounds through a brand you own. If you want an asset at the end, start on a path that builds one.

Most people reading this are best served by wholesale first: real invoices, real suppliers, products that already sell, and an education in Amazon's fees and logistics paid for by goods that move. The rest of this lesson assumes wholesale or arbitrage — you are choosing an existing product to sell. Private label sourcing has its own lesson later.

Step 2. Choose the category with the fee table open

Categories differ in referral fee, in how heavily they are restricted, and in how seasonal they are. Amazon's referral fees run from 5% to 45%, with 15% covering many of the common categories and a minimum of $0.30 on most.¹ A five-percentage-point difference in referral fee is the whole margin on a thin product.

Three category-level questions, before any individual product:

• What does Amazon keep? Look it up rather than remembering it. The tiered categories — clothing, for instance — change rate by price point.
• Is the category gated? Some categories and many individual brands require approval before you can list. Approval usually rests on invoices from a real distributor, which arbitrage receipts do not satisfy. Amazon publishes the current list and the request path in Seller Central; check it before you buy, not after.²
• When does it sell? A product that earns everything in November ties up your cash for ten months and punishes you with long-term storage costs. For a first product, boring and year-round beats exciting and seasonal.

Step 3. Judge the physical object

Fulfilment fees follow size and weight while referral fees follow price, so the shape of the thing decides more than beginners expect. Reject on any of these before you look at demand:

• Heavy or bulky for its price. The classic beginner's mistake: a $20 item that costs a large share of its price to store and ship.
• Fragile. Breakage is paid twice — the refund and the lost unit — and it drags your account metrics down with it.
• Batteries, liquids, aerosols, flammables. These fall under Amazon's dangerous goods handling and add review steps and restrictions. Not impossible, wrong for a first product.
• Expiry dates. Food and supplements come with shelf-life rules and removal risk on top of the usual restrictions.
• Many variations. Twelve sizes in five colours is sixty stock decisions and sixty ways to end up with dead units.
• Products where the buyer needs it to be right. Electronics accessories with compatibility lists, replacement parts by model number: high return rates, and returns are the quiet killer of thin margins.

Step 4. Check what you are allowed to sell

Two different walls sit in front of a listing, and people confuse them constantly.

Category and brand gating is Amazon deciding you need approval. It is usually solvable with the right paperwork from a legitimate supplier.

Intellectual property is the brand deciding you are not welcome. A brand can file a complaint against your listing, and a pattern of complaints is one of the fastest ways to lose an account. Selling authentic goods you bought legally is not automatically a defence if you cannot document where they came from. This is the single strongest argument for wholesale over arbitrage: an invoice from an authorized distributor is a document you can hand over.

Step 5. Read the demand, and the competition

You are trying to answer two questions: does this sell steadily, and is there room for one more seller?

• Sales rank over time, not today. A single rank is a snapshot; a year of rank history tells you whether demand is stable, seasonal or dying.
• How many sellers are already on the listing. On a wholesale product you are sharing one listing with everyone else who buys from the same distributor. Fifteen sellers on a listing means the price will be competed down to the floor.
• Whether Amazon itself sells it. When Amazon is on the listing, it is a difficult listing to win consistently.
• Review count on the top listings. For private label thinking, thousands of reviews on every result is a category that costs money to enter.
• Price stability. A price that swings weekly is a listing where somebody is dumping stock, and your margin calculation is fiction.

Tools help here, and we name them without any commercial relationship: Keepa for price and rank history, Helium 10 and Jungle Scout for research suites, SmartScout for looking at brands and categories from the top down. Golden River Stone has no affiliation with any of them and earns nothing if you subscribe. Start on free tiers. Amazon's own best-seller and movers-and-shakers pages cost nothing and show you real demand.

Step 6. Run the unit economics

Now the arithmetic, on the survivors only. One sheet per product, and every line filled in before you judge it.

Two lines people leave out, and both change decisions. Storage is charged monthly on volume and rises sharply for inventory that sits past six months, so a slow seller loses money quietly for a long time.⁴ Income tax is not in the table above: profit per unit is not take-home.

Judge on two numbers rather than one. Profit per unit tells you whether the product is worth handling at all. Return on cash tells you how fast your money comes back, and with a small budget the speed of the cycle matters more than the size of the margin.

Step 7. Check the cash cycle before you commit

The arithmetic above is per unit and instantaneous. Your bank account is neither.

Mistakes that cost people their first order

• Modelling with zero advertising and zero returns. Both are real, both are per unit, both turn a 20% margin into a 5% one.
• Using today's price as the price. Check what the listing has held over months. Your competitor's clearance is not your business model.
• Buying before checking whether you can list. Approval is requested with an invoice from a legitimate supplier — which means checking the gate before you spend, not after.
• Falling for a big margin on a slow seller. Forty percent on something that sells twice a month is worse than fifteen on something that sells daily, once storage and your time are counted.
• Ignoring how many sellers share the listing. A crowded listing is a price war you will join at the worst moment.
• Spending everything on the first order. If the second order needs money you do not have until Amazon pays out, you have designed a business that stalls the moment it works.

Next

You know what you want to sell. Lesson 2 builds the thing that is allowed to buy it: the company, the tax number, and the annual filings that catch foreign owners by surprise — including one whose penalty starts at $25,000.