Amazon Product Research: The Free Tool and the Numbers That Decide
Amazon product research answers one question: does this item clear a real margin at your price? Start inside Seller Central with Product Opportunity Explorer, free on a Professional selling account. Size demand and competition there. Then build a landed cost model using Amazon’s current referral rate, your own fulfilment figure from the FBA Revenue Calculator, plus returns and advertising. The bottom line decides, not the trend chart.
The question this research answers
Most guides ask what is selling right now. The useful question is narrower.
Can you buy this item, land it, list it, advertise it, absorb returns, and still keep enough per unit to pay for the next purchase order?
Demand data proves a market exists. It never proves you can enter it and keep money. Arithmetic does that.
So split the work in two. Build a shortlist, then push every candidate through a cost model. Most die at the second step, which is what a good process should do.
Start with Product Opportunity Explorer
Product Opportunity Explorer lives inside Seller Central. It costs nothing beyond the Professional selling plan at $39.99 a month, and it does not need Brand Registry. Almost no product research guide mentions it, because no software vendor gains from naming a free first party tool.
Amazon defines a niche as a set of customer needs, built by grouping search terms according to the products shoppers view or buy after searching. It defines unmet demand as collections of similar search terms with lower conversion rates than the benchmark. That gap points at demand which current listings answer badly.
Amazon states that new products launched with insights from the tool show 2.5 times higher sales potential in their first three months. That figure comes from 2025 internal data across US, EU and Japan sellers. The same page calls the tool a guide, not a replacement for your judgement. Read both statements on Amazon’s Product Opportunity Explorer page.
Per niche you get search volume, click share by product, price segments, top products, launch counts, and 12 month demand curves. That answers four questions. Does demand exist, is it growing, who owns the clicks, and at what price do shoppers buy. Pair the search terms it surfaces with your own keyword research before you trust a niche boundary.
Two honest limits. Amazon publishes no Selling Partner API for this tool, so nothing pulls its data into a spreadsheet on a schedule. Sellers also complain in Amazon’s own forums about the keyword grouping, where one niche mixes search terms with different buying intent. Read the niche as a first draft, then check it by hand.
Amazon’s public page does not state a refresh cadence either. Read the date range printed in the tool rather than assuming it updates daily.
The six numbers that decide
Six figures carry most of the decision. The thresholds below are our operating rules after seven years of managing marketplace accounts, not Amazon standards. Tighten them for your own cash position.
| Metric | Where to find it | Green | Amber | Red |
|---|---|---|---|---|
| Niche demand | POE niche search volume, 12 month view | Flat or rising | Falling under 10% | Falling over 25% |
| Click concentration | POE products tab, click share | Top 5 under 50% | 50% to 70% | Over 70% |
| Price band | POE price segments | Your price sits in the largest band | One band away | Outside every busy band |
| Review barrier | Median reviews, top 10 offers | Under 300 | 300 to 1,000 | Over 1,000 |
| Return rate | Category norms, then Voice of the Customer once you sell | Under 5% | 5% to 10% | Over 10% |
| Seasonality | POE 12 month demand curve | Peak under 2x median month | 2x to 3x | Over 3x |
One red kills a candidate. Two ambers usually kill it too, because ambers compound. A high review barrier plus heavy click concentration means you pay for every sale for a long time.
Competition, measured properly
Count the products taking 80% of the clicks in the niche. That number tells you more than any competition score a tool invents.
Four products taking 80% of clicks is a closed niche. Twenty products sharing the same 80% is an open one. POE gives click share by product, so you can add the column yourself.
Then check three more things. Brand diversity, because ten listings owned by two brands is one competitor wearing hats. Median review count across the top ten offers, since that is the real entry price. Sponsored Products usage on page one, because heavy coverage previews your advertising bill.
Review count is the barrier most guides treat as soft. It is not. Reviews accumulate with time and volume, and you cannot buy either. Against a field holding 2,000 reviews each, your path runs through a narrower search term set and a sharper offer. That means a different Amazon product launch plan and a bigger budget.
Landed cost, worked all the way through
Every cost line has to sit in one model. Unit cost, freight, duty, prep, inbound placement, referral fee, fulfilment fee, storage, returns and advertising.
Two rules on the fee inputs. Take the referral rate from Amazon’s pricing page for your exact category. Take the per unit fulfilment fee from the FBA Revenue Calculator in Seller Central. Amazon’s rate card sits behind a login, and third party blogs copy old versions of it.
The 2026 change matters for your model. Amazon announced it on 15 October 2025, effective 15 January 2026, giving sellers at least 90 days notice. Amazon’s own figure: FBA fees rise by an average of $0.08 per unit sold, or less than 0.5% of an average item’s selling price. Amazon states there will be no new FBA fee types in 2026, and the announcement names no US referral fee increase. Read it at Amazon’s 2026 fee update.
The example below is an invented product for illustration. It is a stainless steel pour over kettle in Home and Kitchen, priced at $34.95, at the real 15% referral rate. The $5.00 fulfilment figure is our assumption, used only to finish the arithmetic. It is not an Amazon rate, so replace it with your calculator output.
| Cost line | Example kettle | Your product |
|---|---|---|
| Sale price | $34.95 | |
| Unit cost from supplier | $6.80 | |
| Freight and duty | $1.60 | |
| Prep and labelling | $0.45 | |
| Inbound placement | $0.35 | |
| Referral fee at 15% | $5.24 | |
| FBA fulfilment fee (assumed) | $5.00 | |
| Storage per unit sold | $0.22 | |
| Returns allowance at 4% | $1.40 | |
| Advertising at 12% of price | $4.19 | |
| Contribution per unit | $9.70 | |
| Contribution margin | 27.8% |
Before advertising, contribution is $13.89, or 39.7% of price. Supplier terms and freight route move those first two lines most. That is why sourcing products belongs in the research phase, not after it.
Referral fees by category
Rates below come straight from sell.amazon.com/pricing. The minimum referral fee is $0.30 for most categories, with exceptions Amazon lists on that same page. Check your category there before you model, because price band splits change the answer.
| Category | Referral rate | Notes |
|---|---|---|
| Amazon Device Accessories | 45% | The highest rate on the schedule |
| Appliances, compact | 15% to $300, then 8% | The lower rate starts above $300 |
| Automotive and Powersports | 12% | Fitment data drives conversion here |
| Baby Products | 8% to $10, then 15% | Most listings sit above $10 |
| Backpacks, Handbags and Luggage | 15% | Flat rate |
| Beauty, Health and Personal Care | 8% to $10, then 15% | Pricing at $9.99 saves 7 points |
| Clothing and Accessories | 5% to $15, 10% to $20, 17% above $20 | Three bands, so $21 costs you badly |
| Computers | 8% | Low rate, thin retail margins |
| Consumer Electronics | 8% | Low rate, high return risk |
| Electronics Accessories | 15% to $100, then 8% | Most private label units sit at 15% |
| Grocery and Gourmet | 8% to $15, then 15% | Minimum fee rules differ here |
| Home and Kitchen | 15% | The default private label category |
| Jewelry | 20% to $250, then 5% | The 20% band covers most listings |
| Pet Supplies | 15%, 22% for veterinary diets | Check the sub category before you model |
| Toys and Games | 15% | Seasonality risk sits highest here |
| Everything Else | 15% | The catch all rate |
The margin threshold we hold to
Our operating rule is 30% contribution margin before advertising. That is our rule, not an Amazon benchmark and not an industry average.
Here is the reasoning. Advertising takes roughly 12 points of price on a product that still needs paid traffic to hold rank. Returns and reimbursement gaps take another 3 to 4 points in most categories. That drops a 30% product to about 18%.
Then price erosion arrives. Three competitors land in the niche and someone discounts 10%. Percentage fees fall with the price, but your unit cost, freight and fulfilment fee do not. A product starting at 30% pre advertising lands near 12% after that cut. A product starting at 22% lands close to break even.
The example kettle starts at 39.7% before advertising. After a 10% price cut it still returns 23.1% pre advertising and stays healthy. That headroom is what you are buying when you hold the line at 30%.
Below 30%, the product can still work in two situations. You own a cost advantage no competitor can copy quickly. Or you sell it as part of a range where one hero product carries the advertising. Outside those two cases, a thin margin at launch becomes a loss the first time someone discounts.
Compliance checks before you commit
Do these before the container ships. Fixing a compliance problem with 2,000 units already in a warehouse costs far more than a week of checking.
Category approval comes first. Amazon restricts several categories and requires an application, and the requirement depends on your account and supplier documents. The Selling Applications page in Seller Central shows your status. No third party tool can see it.
Hazmat classification catches sellers out. Batteries, aerosols, magnets, flammable liquids and some cosmetics trigger a dangerous goods review. That delays your first shipment and can limit fulfilment options.
Brand gating is separate again. Some brands and some ASINs need approval even inside an open category, so a reseller model needs invoices ready.
Then GTIN. Amazon matches product identifiers against the GS1 database, so buy barcodes from GS1, not a reseller. A rejected GTIN blocks the listing on launch day.
Free tools against paid tools
Pay for the jobs Amazon does not cover, and stop paying for the ones it does.
| Job | Amazon free tool | Paid alternative | Worth paying |
|---|---|---|---|
| Find niches with unmet demand | Product Opportunity Explorer | Vendor niche finders | No, the first party data wins |
| Estimate one ASIN’s sales | POE product view | Vendor sales estimators | Sometimes, for single ASIN checks |
| Search term volume and share | Brand Analytics, brand owners only | Vendor keyword tools | Yes, if you have no Brand Registry |
| Fee and profit maths | FBA Revenue Calculator | Vendor profit calculators | No, use the source |
| Price and rank history | Nothing published | Keepa and similar | Yes, history exposes price wars |
| Supplier discovery | Nothing published | Vendor supplier databases | Sometimes, verify every match |
One subscription plus Product Opportunity Explorer covers a first product properly. Three subscriptions rarely change a single decision.
Where amazon product research goes wrong
Five failure patterns account for most of the money lost before launch.
Choosing on demand alone. A big niche with four dominant sellers is worse than a small niche with twenty.
Ignoring returns. Apparel and electronics carry return rates that quietly erase a spreadsheet margin. One returned unit costs you the outbound fee, the return processing and often the resale value.
Underrating the review barrier. Reviews take time, and time costs storage fees and interest on stock.
Modelling on today’s price. Model the price after three competitors arrive, because they will.
Forgetting that stock is working capital. A 20% margin turning six times a year beats a 27.8% margin turning twice. Cash cycle belongs in the model.
Best Sellers Rank deserves a warning too. Amazon states BSR reflects sales rank against similar products, calculated from sales volume, with recent sales counting more than older sales. Its seller facing page gives no update frequency, so treat any tool claiming an hourly cadence with care. Read Amazon’s own explanation of Best Sellers Rank.
How we run this for a client
The inputs are short. Your target category, your capital, your supplier options and your risk tolerance.
The work is a shortlist from Product Opportunity Explorer, a click concentration count per niche, and a review barrier count. Then a full landed cost model per candidate, built on your real supplier quotes.
The output is a decision document. Each candidate gets its six metric grid, its margin at launch price, its margin after a 10% price cut, and its working capital need.
The choice stays with you. We do not promise a ranking, a Buy Box share or a sales figure. We have managed over $100M in marketplace sales since 2019, and good candidates still fail for reasons no spreadsheet catches.
Frequently asked questions
How do I research a product to sell on Amazon?
Open Product Opportunity Explorer in Seller Central and find niches with unmet demand. Check click concentration, price bands and review counts in the top ten offers. Then build a landed cost model with your category referral rate and your Revenue Calculator fulfilment fee. Keep only candidates that clear your margin rule after returns and advertising.
Is Product Opportunity Explorer free?
Yes. It comes with a Professional selling account at $39.99 a month plus selling fees, and it does not require Brand Registry. There is no separate charge for the tool. Amazon publishes no Selling Partner API for it, so you cannot automate exports, and you read the data inside Seller Central.
What is a good profit margin for an Amazon product?
We hold to 30% contribution before advertising. Advertising takes around 12 points on a product needing paid traffic, and returns take 3 to 4 more, so 30% lands near 18% net. A 10% price cut from a new competitor then drops it near 12%. That is our operating rule, not an Amazon benchmark.
How much do Amazon referral fees cost?
Most categories charge 15%, with 8% categories such as Computers and Consumer Electronics, and 45% on Amazon Device Accessories. Several categories split by price band, including Clothing and Accessories at 5%, 10% and 17%. The minimum referral fee is $0.30 for most categories. Check sell.amazon.com/pricing for your exact category.
Did Amazon fees increase in 2026?
FBA fees rose by an average of $0.08 per unit sold from 15 January 2026, on Amazon’s own figure. Amazon describes that as less than 0.5% of an average item’s selling price. It announced the change on 15 October 2025 with at least 90 days notice, added no new FBA fee types, and named no US referral fee increase.
How many reviews do competitors have before a niche is closed?
There is no published cutoff. Our rule uses the median review count of the top ten offers. Under 300 is workable, 300 to 1,000 needs a sharper angle and a bigger budget, and over 1,000 usually means years of catching up. Pair that count with click concentration before you decide.
What is a good BSR to target?
Amazon states BSR reflects sales rank against similar products, calculated from sales volume, with recent sales weighted above all time sales. Amazon publishes no conversion from rank to units, and ranks are not comparable across categories. Use BSR to compare products inside one category only, and size demand from search volume instead.
Are free product research tools good enough?
For a first product, yes. Product Opportunity Explorer and the FBA Revenue Calculator cover niche discovery and unit economics with Amazon’s own data. Paid tools earn their fee on price and rank history, and on keyword volume if you have no Brand Registry. Buy one subscription, not three, and only after your shortlist exists.
How do I check whether a category is gated?
Check inside your own Seller Central account, because gating depends on your account history and your documents. Search the product in Add a Product and look for an apply to sell prompt, then open Selling Applications to see requirements. A category open to one seller can stay closed to another, so no third party tool can answer this.
How much capital do I need to launch one product?
Amazon publishes no figure, and any number you see quoted is someone’s guess. Model it instead. Add two purchase orders of stock, freight and duty, prep, your first 90 days of advertising, and the cash gap between paying your supplier and Amazon paying you. That total is your real requirement.
How do I account for returns in my model?
Put a returns allowance in as a percentage of price, and set it from your category rather than a single average. A returned unit costs you the outbound fulfilment fee, return processing and often the resale value, so the true cost sits above the refund. Once you sell, replace the estimate with Voice of the Customer data.
Should I use Helium 10 or Jungle Scout for product research?
Either works, and neither replaces Product Opportunity Explorer, which holds Amazon’s first party click and conversion data. Pick one on the job you need most, usually keyword volume or historical price and rank. Trial both against a niche you already know well, then keep whichever matched reality more closely.
Want a second opinion on your shortlist
Send us the three products you are weighing up and the supplier quotes behind them. We will run the click concentration count and the landed cost model, then tell you which survives. Our Amazon account management work starts with a free audit, no setup fee and no lock in.
