Most people who fail on Amazon make their fatal mistake before they order a single unit. Not in their launch, not in their ads. In product selection. And it almost never feels like a mistake while it is happening. It feels like excitement.
That is the trap. You find a product you like, the top-line revenue looks strong, and every instinct says go. Then the referral fees, the FBA costs, the returns, and the ad spend it takes to stay visible quietly eat the margin, and you are left holding a warehouse full of a decision that never had the numbers to work.
Good Amazon product research is the discipline of running that math before you commit, and letting it outrank how much you like the product. This post walks through how to do it: how to read a market, how to judge real demand, where the best opportunities actually hide, the numbers that decide, and the one trap that catches almost everyone. At the end, you get the full checklist we run on every product before a dollar moves.
Start before the product: your budget and your goal
Product research does not start with the product. It starts with two questions that have nothing to do with it.
What can you actually invest, and what is your timeline? That answer sets the price point you can play in. A $25 product and a $65 product are different businesses, and part of your budget has to stay reserved for advertising at launch, not just inventory. If you spend everything on units and have nothing left for ads, the product never gets seen.
Timing matters too. A product that sells heavily in December can be irrelevant by February. A seasonal product can be a perfectly good decision, as long as you know that is what you are buying. It means ordering higher quantities than average and carrying a heavier ad budget through the window. The mistake is backing a seasonal product without realizing it, then wondering why sales fell off a cliff.
Read the niche: competition you can actually beat
The next question is whether you can enter this market at all. Most sellers get this wrong in one of two directions.
They pick a niche that is too small, with no real demand to capture. Or they pick one that looks busy and mistake it for healthy, when in fact one dominant brand owns it and no newcomer has broken in for years. The skill is telling the difference between a dead niche and a mature one.
A few things to look at:
- The selling price range in the niche, so you know where your economics would land.
- The number of competitors, and more importantly how demand is distributed across them.
- The review-count range, which tells you how entrenched the leaders are.
- Whether one brand holds most of the market.
If a single seller owns more than roughly 70% of the category, that is usually a signal to walk away. You would be spending your entire budget trying to buy your way in against an incumbent, and that rarely ends well.
There is also a judgment call about complexity. Do you want a complex product with many ways to go wrong in manufacturing and returns, or a simple one that is harder to differentiate but easier to run? Neither is wrong. But decide it deliberately, and check compliance requirements before you get attached to anything.
The tip most people miss
Here is a fast way to know whether a niche is actually open: find a recent launch in it. A listing with under 100 reviews went live only a few months ago. Look at how many units it is moving. If someone just succeeded in this niche from a standing start, that is your proof the door is open and you can too. A dead niche does not produce new winners. If the only successful listings are years old with thousands of reviews, that tells you something different.
Read the reviews, but not the good ones
The best product opportunities hide in competitor reviews. Not the five-star ones, the one and two-star ones.
Five-star reviews tell you what is already working, which you cannot easily improve on. One and two-star reviews tell you what customers are consistently not getting. That gap is where you enter. If every negative review of the category leader complains about the same thing, and it is something you can fix, you have found your angle.
The key word is consistent. One angry review is noise. The same complaint appearing again and again, across multiple competitors, is a pattern, and a pattern you can solve is a real reason to enter a market. It also gives you your differentiation: a genuine added value you can show in your main image, not a me-too listing that competes on price alone.
The numbers that decide
This is the heart of product research, and it is where excitement goes to die.
A product can clear every test above and still be a no, because the margin does not survive the real cost of selling on Amazon. The number that matters is not the price, and not the top-line revenue. It is what is left after everything Amazon and the supply chain take out.
Run every serious candidate through the full cost stack:
What is left is your contribution margin. And here is the rule that matters most: a margin that looks thin before you have spent a dollar only gets worse. As competition pushes ad costs up, single-digit margins do not survive. You cannot optimize your way out of a margin that never worked on paper.
So what threshold should clear? A useful general target is 20%. On a cheaper product, hold firm above it, because there is no dollar cushion to absorb a bad month. On a more expensive product, sometimes 17% is acceptable, but only once you have honestly accounted for how long a tough niche will keep your ad spend high and your margin suppressed. If you are entering a hard category, you may be spending heavily on ads for a long time before the margin recovers, and that has to be in the math from the start.
A product that comes in at 8% is not a product you fix. It is a product you pass on.
A quick worked example
To make it concrete, picture two products you might evaluate in the same week.
The first looks like an obvious winner. Strong estimated revenue, a category that is clearly growing, and a gross margin that looks fine at a glance. Every gut signal says go. But run the full stack: the referral fee, FBA, returns, and the ad spend needed to hold rank, and the contribution margin lands in single digits before a single unit ships. One dominant brand already owns most of the category’s reviews. You would be buying your way in against an incumbent on a margin that only shrinks. It is a no, no matter how exciting it looked.
The second is less thrilling. A thinner top-line, a less interesting category. But demand is spread across a dozen sellers rather than owned by one. The negative reviews of the incumbents all complain about the same fixable thing. And once you run the numbers, the all-in margin clears your threshold with room to spare.
Less interesting. Far more viable. That trade, boring and viable over exciting and thin, is most of the job.
The “I love this” trap
Which brings us to the most common reason sellers fail, and the one almost nobody admits to while it is happening.
People choose products because they find them interesting, not because the market demands them. It is a natural pull. You are going to spend months of your life on this product, so of course you gravitate toward one you like. But liking a product is not a market signal. The customer does not care whether you find it interesting. They care whether it solves their problem better than the alternatives.
The discipline is simple to state and hard to practice: you are choosing this product because the numbers say so, not because you like it. When the two conflict, the numbers win. Every time.
The “before you order” checklist
Put every product you are considering through this before you commit:
Points 5 and 7 are the deal-killers. If a product fails either one, it does not matter how good the rest looks. Those two kill more deals than everything else combined.
The takeaway
Amazon product research is not about finding a product you love. It is about running an honest calculation before you spend, and having the discipline to walk away when the math says no, even when everything else says yes. The sellers who last are not the ones with the best instincts. They are the ones who let the numbers outrank their instincts.
We publish the real teardowns behind posts like this in our newsletter: actual products we evaluated, the real numbers, and the decisions we made, including the ones we got wrong. Subscribe below to get the next one.