How to Pick a Profitable Niche Before You Write Anything

Watches pay 4.00% in the United States. The same category pays 6.0% in the United Kingdom and 6.0% in Germany. Luxury beauty runs the other way, 10.00% in the American store and 6.0% in the other two. Two categories, three countries, and the order between them flips with where the buyer stands.

Which niche to choose is a question this site already answers, and the shortlist lives in the niche roundup. This page is the arithmetic underneath it: the inputs you can look up before a single article exists.

I run Amazon Associates in all three of those markets and read all three rate cards side by side on August 21, 2026. One correction first. One older address on the American site redirects to a sign-in screen. The standard commission income statement page does not, and it opened in all three stores with no account.

A category is not one rate, it is three

Each store publishes its own card. The three read side by side look like this.

Two things there change how a shortlist reads. The first is the bottom row, which reverses the pattern above it: a category with no line of its own is worth more in the American store than in the European ones. Televisions, toys, pet supplies and headphones have American rows and no British or German ones, so on those cards they fall to that default. It is a reading of the card, not a line Amazon prints.

The second is the rows paying nothing, which differ by store. Wine has its own excluded line on the British and German cards, while the American one folds it into a broader alcohol exclusion. Check the exclusions on every card you sell into. The full American ladder is in commission rates by category.

Amazon decides which category your product sits in, not you

The British and German appendix carries a definition the American page does not. A product category is the category of the product described in its tables, in Amazon's words "as determined by us", and the same passage adds that Amazon will determine the classification of products in each category.

A wearable you filed under Uhren at 6.0% may be classified as Mobile Elektronik at 2.5%. A countertop machine you called kitchen may land in the appliance row at 2.5% rather than the kitchen row at 5.0%. No Amazon page shows the assignment before you have referred an order, so the planned rate is an assumption until orders arrive.

So if your candidate sits near a boundary, price it at the lower row and see whether the case still holds. A niche that only works at the higher number rests on a decision somebody else makes after you have built.

The commission follows the basket, not the link

The same appendix defines two kinds of qualifying purchase. A direct one comes from the same category as the detail page your link pointed at. An indirect one comes from a different category and pays at that category's rate, not yours. A special link that does not reach a product detail page makes everything arising from it indirect.

The American statement carries neither definition, so the distinction belongs to the British and German appendix and does not generalise to the American store.

That cuts both ways. A modest row is not the ceiling, because a basket can carry categories you never linked, and a high row is not banked.

The window is tight. A session ends 24 hours after the click, when the customer orders a product that is not digital, or when they click another associate's special link, whichever is first. A basket item has to be ordered within 89 days of that click, then shipped and paid within 180 days. A niche people research for weeks loses sessions a short-path niche keeps.

Data freshness is a property of the niche, not of the site

Two rules govern the numbers on a comparison page, both in the participation requirements.

• Price and availability may be shown only when Amazon supplies the link, or when the data comes from the Creators API or the Product Advertising API under its license terms. A price typed by hand is outside the rule.
• Product advertising content that is not an image may be held for at most 24 hours, then refreshed by a fresh call. Images may not be held at all, only the link to one, for up to 24 hours. ASINs may be stored indefinitely.

Multiply that by the catalogue. Two hundred products is two hundred refreshes a day with no season off. That is not writing work. It is a scheduled operation with retries, failure handling and quota accounting.

Variant density is where the niche writes its own bill. One watch, to a reader, is one watch. On Amazon it is strap, dial and case size, each combination its own ASIN, and clothing and footwear split by size and colour the same way. Two niches with the same page count can sit far apart on maintenance.

There is a sequencing problem underneath. The data layer behind comparison pages opens only to an account already making qualifying sales, so it is shut while the site is weakest. The API write-up covers that gate.

A rate you cannot reach is not a rate

Three markets means three tags, three cards and three catalogues, not one catalogue translated. The German 6.0% on watches is worth nothing if the products you cover are not listed on the German store. Search both European stores for ten candidate products and count how many come back before treating a European rate as real. How one site addresses several stores is covered in running one site across countries.

The 180 day clock rewards buying intent, not interest

The clock starts when you apply, not when you are approved. Amazon's dormancy page is blunt: an application can be withdrawn when there have not been enough qualifying sales in the 180 days after submission, and a rejected account or associates ID is not reinstated. Applying again once your site is established is allowed, but the identity does not return.

My American account was closed under that rule once, which is why I read the clock as a niche input rather than an administrative step. The category has to produce three qualifying sales while the domain is newest and least visible, a different test from whether people enjoy reading about it. The mechanics are in the 180 day rule.

The method, input by input

Together they make a short list, and the last column matters as much as the first.

What the decision costs to run afterwards

None of that is the expensive part. The expensive part repeats.

• A daily refresh of every non-image value, sized by product count rather than page count.
• Image handling as a separate recurring job, because images cannot be held and only their links can.
• Three catalogues if you sell into three stores, each with its own tag and reporting.
• A category assignment you cannot preview, which can move a rate underneath your plan.
• A rate ladder Amazon can reprice whether or not you act, which is why one high row is a fragile foundation.

The weekly shape of that is measured in hours per week. Choosing a niche is a thinking problem and everything after it is an operating problem, not the same kind of work.

Where usebravery fits, and where it does not

usebravery is the operating half. You choose the category and decide which products go in, the site is built and hosted, and price and availability are kept refreshed on the schedule the rules demand, with each market running on your own Amazon Associates tag so the commission is paid to you and we take no share of it. The plan question then turns out to be the niche question again. Starter covers the American store, Plus adds the British one and Pro adds Germany, so if your candidate pays 4.00% in the United States and 6.0% in Europe, the market count is a niche decision before it is a billing one. Market count and language count are separate settings, and the plan page sets both out.

What it does not do is make a wrong category right. A niche in an excluded row pays nothing however well the site is run, and a catalogue that does not exist on the German store will not appear because a plan includes Germany. The choice stays yours.

Your next step, in the order it has to happen

In this order it takes an evening rather than a month of circling.

• Write the candidate category down in plain words, then find it on all three cards.
• If it has no row of its own, accept the default, 4.00% in the United States and 3.0% in the other two.
• Read the excluded list on each card and confirm no part of the catalogue sits inside it.
• Open ten listings and count the variants behind them, because that count is your daily refresh bill.
• Search both European stores for those same ten products and see how many exist at all.
• Split your query list into reading and buying questions, then ask whether the buying half can produce three sales while the domain is new.
• Decide the market count last, because by then it is arithmetic rather than preference, and the plans follow from it.

I am Serdar D., founder of usebravery, and I run the program in three markets. One note on why there is no earnings figure above. Amazon's own agreement says it makes no representation, warranty or covenant about the traffic or commission income you can expect. If the company paying the commission will not forecast it, nobody handing you a niche list should either.