How to find products to sell on Amazon: a research method that works

Every brand I have launched started with the same question. Not how to run ads, not which supplier to use, but what to sell. Product research is the part that decides everything downstream, and it is also the part most people rush. They fall for an exciting idea, order stock, and learn the hard way that nobody was searching for it, or that the margin never existed in the first place.

I have spent years picking products for the brands I manage, and the method has barely changed. It is less about finding a magic product and more about disqualifying the bad ones quickly until a good one is left standing. This guide is that method, the criteria I screen for, where I look for ideas, how I check demand and competition, and the traps that quietly eat months. No guru product lists, no get-rich promises, just the process I actually use.

What product research actually is

The mistake most beginners make is treating product research like a treasure hunt, as if there is one hidden product that prints money and the only job is to find it. There is not. Product research is the opposite. You start with a long list of plausible ideas and you remove the ones that fail, one filter at a time, until only the few that survive every check are left.

The numbers above are illustrative, but the shape is real. You will look at far more ideas than you order. An idea that excites you but has no search demand is gone. One with demand but a page one full of thousand-review giants is gone. One that survives both but loses money after fees is gone. By the time you reach a product worth ordering, you have rejected most of what you started with, and that discipline is what separates a launch that works from a warehouse full of dead stock.

Once you see research this way, the pressure changes. You are not trying to fall in love with one idea, you are trying to find reasons to reject each one. The good product is simply the one that refuses to be rejected.

The six things that make a product worth selling

A product earns an order when it clears six things together, not one in isolation. I weigh them as a set, because a product can look brilliant on one and quietly fail on another, and the failure is the part that costs you.

Demand that is real and provable. People have to be searching for it or buying it already. I want a main keyword with steady search volume and top listings that each move a few hundred to a couple of thousand units a month. No demand means you are paying to teach the market that a product exists, which is the slowest and most expensive way to launch anything.

Competition you can beat. Strong demand is worthless if page one is locked down. I look at how many reviews the leaders carry and how good their listings are. A page one averaging a few hundred reviews is catchable in a reasonable time. One where everyone sits above a thousand reviews with polished images is a wall, not a door.

A margin that survives every fee. This is the criterion beginners skip. After the landed unit cost, the freight, the Amazon referral fee, the FBA fee, and a realistic ad budget, the product still has to leave roughly a third. A healthy-looking price means nothing until you subtract all of that. Run it through the fee math before you trust any margin.

Room to differentiate. If the only thing you can change is the logo, you are launching a clone, and the only lever left is price, a race you lose to whoever has deeper pockets. I want at least one real improvement I can make, a fix to a recurring complaint, a better material, a smart bundle, a sharper use case.

Sensible size and weight. A small, light, sturdy product keeps FBA fees low and breakage rare. Heavy, bulky, or fragile products bleed margin into fulfilment and returns before you have sold a single unit.

No regulated or seasonal traps. I avoid electronics, supplements, anything that touches skin or food, and products that only sell three months a year. The approvals, the liability, and the dead months are not worth the headache on a first product, when your job is to learn the business, not to fight it.

A product that clears all six at once is genuinely rare, and that is the whole point of treating them as a set. Most ideas score well on two or three and fail the rest, and a single hard failure is enough to pass on the idea. I am not looking for a product I can talk myself into, I am looking for the one that survives all six honest questions without special pleading.

Where good product ideas come from

Ideas are not really the bottleneck, good ones are. The trick is to look where demand reveals itself instead of brainstorming in a vacuum and hoping. These are the sources I actually use, day to day.

The best ideas usually come from combining two of these at once. A complaint that shows up again and again in the negative reviews of a best-selling product, for example, is both proven demand and a built-in way to differentiate, which is as close to a green light as research gets. I keep a running list as I browse and I do not judge entries yet. The filtering comes later. At the idea stage the only job is to gather enough candidates that you can afford to be ruthless when the cutting starts.

How to check the demand is real

An idea feels like demand. Data proves it. Before a product goes any further, I confirm that people are genuinely looking for it and buying it, in numbers, not that it merely seems like something people might want.

The first signal is search volume on the main keywords. If a few hundred to a few thousand people a month search for the product, there is a market to capture. If almost nobody searches, you would be relying entirely on browse traffic and ads to create demand from scratch, which is slow and expensive for a brand with no history.

The second signal is how much the existing listings sell. Amazon hides exact sales, but the Best Sellers Rank is a usable proxy, and the lower the rank number within a category, the more that product moves. I want to see several listings, not just one, holding steady volume. A single dominant listing with a long tail of dead ones is a one-winner category, and the winner is almost never the newcomer.

The third signal is depth. A healthy product has demand spread across several related keywords, not one fragile term. That spread means you can rank for a cluster and survive losing any single phrase. I would rather have steady demand across ten ways of describing a product than a spike on one keyword that a competitor can outbid me on tomorrow morning.

One warning on demand. High search volume on a single broad keyword can be a mirage if that keyword is generic and fought over by huge brands. A thousand searches on a tight, specific phrase that a buyer types when they are ready to purchase is worth more than ten thousand on a vague term that browsers type with no intention of buying. I read the intent behind the demand, not just the size of it.

How to read the competition

Demand tells you the room is full of buyers. Competition tells you whether you can actually reach them. The two have to be read together, which is exactly what the matrix further down is built for.

I judge competition on three things. The first is reviews, because reviews are the moat. Catching a listing with two hundred reviews is a few months of honest work. Catching one with five thousand is a different sport, and usually not one worth entering as a first product. The second is listing quality. If the top sellers have lazy images, thin bullets, and no real optimised listing, there is room to win on execution alone, even in a busy category. The third is brand dominance. A category owned by one strong brand with loyal buyers is far harder than one split among ten mediocre sellers, even when the review counts look similar on paper.

The honest test is simple. Look at page one and ask whether you could produce a clearly better listing than at least a few of the results, within your budget. If the answer is yes, the competition is beatable. If every result already looks better than anything you could make, the demand does not matter, because buyers will never get far enough down to see you.

Plot it: the demand and competition matrix

Demand and competition only mean something together. High demand with low competition is the product you want. High demand with fierce competition can still work, but only with real differentiation and a budget to back it. Low demand with low competition is a small but defensible niche. Low demand with high competition is simply a trap. Plot your idea below and see which one you are actually looking at.

Most beginners spend all their time in the top-right corner, chasing the obvious high-demand products that everyone else is chasing too, then wonder why the launch stalls against entrenched sellers. The quiet money is usually in the top-left, demand that is real but that the big sellers have not bothered to serve well. That corner is where a small, sharp brand can still win, and finding it is the entire point of the work above.

Run the money before you fall in love

A product can pass demand and competition and still be a bad business, because the margin does not survive the fees. This is the check that turns an exciting idea into a real one, and it is the one people most want to skip past.

Take the price the market actually charges, not the price you hope to charge, and subtract everything. The unit cost landed at your warehouse, the freight, the Amazon referral fee, the FBA fulfilment fee, and a realistic slice for advertising, which on a brand-new product is rarely small. What is left is your real margin. If that number is not roughly a third of the price, the product is fragile, because the first storage bill, return wave, or price dip will eat straight through it.

I run this math before I order samples, not after. The Amazon fees guide breaks down each cost in detail, and the launch calculator in the private label guide turns your numbers into a breakeven and a profit figure in seconds. A few minutes here prevents the most expensive mistake in the business, ordering a product that was never going to make money once the fees were counted.

What this looks like on one real idea

Theory is easy, so here is the method applied to a single idea, the way I would actually run it. Say the idea is a set of reusable silicone stretch lids, the kind that seal over bowls and tins instead of cling film.

I start with demand. The main keyword and its variants pull a few thousand searches a month between them, and several listings on page one are each moving what looks like a thousand-plus units a month judging by their rank. The demand is real and spread across several phrasings, not balanced on one fragile term, so the idea clears the first filter comfortably.

Then competition. Page one is a mix, two strong listings with several thousand reviews each, and four or five weaker ones in the few-hundred range with mediocre photos and vague bullets. That is a pattern I like. The leaders are strong but they are not the whole page, and the weak listings prove there is room for better execution. It is not an open field, but it is beatable, so it passes the second filter, leaning toward the busier side.

Now the money. The product sells around twenty-two, the landed unit cost is roughly four, Amazon fees take about six, and I pencil in a few more per unit for launch advertising. That still leaves a margin comfortably above a third after ads, so the economics hold and the third filter is passed.

On the matrix this idea sits high on demand and middle-to-high on competition, the top of the chart leaning right. Not the clean opportunity corner, but a workable red-leaning spot, the kind that pays off only if the differentiation is genuine.

So the last question is differentiation. The negative reviews of the leaders complain that the lids slip off larger bowls and tear at the edge. That is the opening, a thicker edge and one extra-large size, led with in the first image and the opening bullet. Now it is not a clone, it is the version that fixes the top complaint buyers already have. That is an idea worth ordering samples for. Notice that the product itself is dull, and that dull but validated beats exciting but unchecked every single time.

Do you actually need product research tools?

Most of page one for “amazon product research” is software companies selling you their tool, so it is worth saying plainly. Tools help, but they are not the method, and you do not need a paid one to begin.

What the tools actually do is speed up the boring parts. They estimate sales from rank, pull search volumes, and let you filter thousands of products by price, reviews, or weight in seconds instead of opening listings one by one. For someone screening a lot of ideas, that time saving is real, and the better ones pay for themselves comfortably once you are ordering stock regularly.

What they do not do is make the decision. A tool will happily hand you a product with great numbers and a fatal flaw it cannot see, a design you cannot improve, a brand that owns the category, a margin that dies after ads. I have watched people order confidently because a tool gave a product a high score, and lose money all the same. Start by checking a dozen ideas by hand, so you learn what the numbers actually mean. Add a tool when the manual work becomes the bottleneck, not before. The judgement stays yours either way, and the judgement is the part that matters.

The methods that waste your time

Some of the most popular product research advice actively wastes your time, and learning to ignore it is half the battle.

The guru product list is the worst offender. The moment a product appears on a public list of winners, every beginner who bought that course orders the same thing, the category floods, and the margin collapses within weeks. By the time a product is famous enough to make a list, the opportunity has already gone. The same applies to the “top ten products to sell this year” articles, which exist to collect clicks, not to make you money.

Blind copying is the next trap. Picking the current best seller and launching an identical version puts you at the back of the line, behind an established listing with thousands of reviews, competing on a price you cannot win. Copying a product without improving it is not research, it is volunteering to come last.

Chasing a trend late is the third. Trends are genuinely useful early, when interest is rising and the category is still thin. Jumping in after a product has already gone viral means arriving exactly as demand peaks and competition explodes, then holding stock as the wave breaks under you. If everyone already knows about it, you are early to nothing.

The mistakes that quietly kill product research

Beyond the bad methods, the same few mistakes sink the research itself, no matter how good your sources are.

Falling in love before the data is the big one. The moment you want a product to work, you start reading the numbers to confirm your hope instead of to test it honestly. I qualify with the data first and decide whether I like the product second, never the other way around.

Checking one factor in isolation is the next. A product with enormous demand still fails if the competition is a wall or the margin is thin. The factors only mean something as a set, which is why I never green-light an idea on the strength of a single good number.

Quitting the search too early is quieter but just as costly. The first idea that looks decent is tempting after hours of looking, but settling for a six-out-of-ten product because you are tired of searching is how mediocre launches are born. The discipline to reject a nearly-good product and keep looking is what eventually turns up the genuinely good one.

What to expect, and the habit that makes it easier

Two things surprise people about real product research. The first is the hit rate. You will reject far more ideas than you keep, and good ones are rare, so checking twenty or thirty ideas to find one worth ordering is normal, not a sign you are doing it wrong. If every idea is sailing through your filters, the filters are too soft.

The second is that it is never really finished. The best operators treat product research as a running habit, not a one-off sprint before a launch. I keep a simple swipe file, a single note where I drop any product that catches my eye while I shop, read reviews, or browse a category, with one line on why. Most entries never go anywhere. But when it is time to find the next product, I am not staring at a blank page under pressure, I am starting from a list of pre-screened candidates gathered calmly over months.

That habit changes the quality of the decision. Research done in a rush, against a deadline, with cash waiting to be spent, is where the worst choices happen, because the pressure pushes you to settle for whatever is in front of you. Research done continuously, with no money on the line yet, lets the data do its job. The single best upgrade you can make to your product research is to start it long before you actually need a product.

And keep the bar fixed as you go. It is tempting to lower your standards after a long search, but the product you settle for on a tired afternoon is the one you live with for a year of inventory. Hold the line, and let the swipe file, not the deadline, decide when you have actually found the one worth ordering.

Where product research fits

Product research is the foundation everything else sits on. Get it right and the rest of the work compounds, the listing optimisation, the FBA logistics, the fee math, and the private label brand you build on top of it. Get it wrong and none of those can rescue a product the market never wanted in the first place.

This is the part I spend the most time on for the brands I run with Novazon, because every later decision inherits whatever the product research got right or wrong. You can see how I work on my Amazon page.