How to find products to sell on Amazon UK and Europe without guessing

1. The short answer

Treat every product as a candidate until it passes eight gates:

1. A specific customer need and a credible demand signal.
2. A competition profile you can actually understand.
3. A plausible and documentable supply path.
4. A contribution model that survives realistic costs.
5. A cash cycle that fits your available working capital.
6. A product-risk screen covering restrictions, compliance and intellectual property.
7. A fulfilment, damage and returns assessment.
8. A written decision with clear stop conditions.

If one of the important gates is unknown, the answer is not “buy a small amount and hope”. The answer is to name the missing evidence and decide whether it is worth obtaining.

For a reusable record, start with the Amazon product research checklist. For a way to compare several candidates, use the Amazon product opportunity scorecard.

2. 1. Start with a customer problem, not a product list

“Kitchen products” is not a product idea.

“A stackable organiser for renters who need to use a narrow under-sink space without drilling” is closer. It gives you a customer, a use case, a physical constraint and questions you can test.

Write the candidate down before opening a research tool. Include:

This is not busywork. An unspecific idea lets every positive signal look relevant. A precise candidate lets you find contradictions.

Amazon's own in-demand product categories page is useful for seeing the categories Amazon is currently promoting and the research and fee tools it points sellers towards. It is Amazon's commercial perspective, though. A category being promoted is not independent proof that your product will be profitable.

3. 2. Gate one: test demand with more than one signal

Demand evidence comes in different strengths. Do not combine every number you see into a confident sentence.

I keep four labels separate:

• Actual first-party data: information available in the seller account or a named marketplace report, with the store and date recorded.
• Relative public signal: a ranking, trend or category snapshot that shows direction or prominence, not units sold.
• Third-party estimate: a tool's modelled estimate. It can be useful for prioritising research, but it is not an Amazon report.
• Assumption: an input you have chosen because you need to model a case. It is not evidence until tested.

Product Opportunity Explorer

Amazon describes Product Opportunity Explorer as a way to explore customer needs grouped into niches. Its public explanation says the tool can be browsed by category, keyword or product and can surface information about demand, purchasing behaviour, competition, search terms, reviews and returns.

That makes it useful for forming questions such as:

• Are customers searching around a recognisable need?
• Do customers appear to be dissatisfied with existing products?
• Is the need seasonal or persistent?
• Is the competition concentrated among a few brands?
• Do returns or review themes point to a product problem rather than an easy opportunity?

The exact fields, access and store behaviour depend on the account and marketplace. Verify the signed-in experience before publishing a screen-by-screen guide. Do not use the public description as evidence that a particular candidate will sell.

Amazon also says the tool is guidance and does not guarantee a particular outcome or sales. That limitation is important. The tool can help you choose what to investigate. It cannot choose what to buy for you.

Best Sellers and public category snapshots

The Amazon UK Best Sellers pages can show which products are prominent in a category at a particular moment. Record the category, product configuration and date if you use the page.

Do not turn a Best Sellers Rank into a unit-sales claim unless you have a disclosed, validated model for that exact marketplace and category. Rank is relative and dynamic. It tells you that a product is prominent in a comparison set. It does not tell you your likely sales, margin or access to supply.

Google Trends

Google Trends can help you inspect interest over time and compare regions. It uses normalised, relative values. It is not absolute search volume and it is not Amazon demand.

When I use it, I record:

• the exact term or topic;
• the country or region;
• the date range;
• the search type;
• whether the result is a term or a topic; and
• the question the comparison is meant to answer.

Use Google's explanation of terms and topics when a product name is ambiguous or translated differently across countries. A falling term can reflect a naming change. A rising term can reflect news, a one-off event or a different meaning.

A simple demand record

If all your evidence says “people are looking”, you have not yet answered whether they will buy your specific configuration at a price that supports your costs.

4. 3. Gate two: inspect the shape of competition

Competition is more than the number of reviews on the first listing.

For a candidate, record the competitive shape on one date:

A high review count can indicate demand. It can also indicate a mature listing that will be difficult to displace. A low review count can indicate a gap. It can also indicate weak demand, poor supply or a product that creates returns.

Keep two questions separate:

1. Is there a customer need that appears underserved?
2. If I source this exact product, can my offer compete with the existing offers?

The first is a product-niche question. The second is an offer and supply question. The detailed Amazon competition analysis guide will own the first. The wholesale ASIN competition guide will own the second for catalogue buyers.

5. 4. Gate three: connect the idea to a credible supply path

Do not contact suppliers with a product name alone.

Create a short product specification first:

• exact dimensions and tolerances;
• material and finish;
• unit, pack and case configuration;
• required labels and packaging;
• target store and customer use;
• known safety or compliance questions; and
• the evidence you need before placing an order.

Then decide which supply route makes sense. A manufacturer, brand distributor, wholesaler and sourcing agent create different evidence questions. The supplier type guide explains that choice. The European supplier discovery guide covers how to build a longlist without relying on one directory.

For each serious candidate, I want to know:

1. Who is the legal entity?
2. What role does it actually play in the supply chain?
3. Can it identify the exact product and configuration?
4. Can it supply a sample or clear specification?
5. Will the quotation, invoice, payment beneficiary and goods agree?
6. What happens if the product, packaging or lead time changes?

The supplier-vetting guide owns the broader identity and payment checks. Do not use “supplier found” as a synonym for “supply route verified”.

6. 5. Gate four: calculate contribution, not just gross margin

The product price is not the product economics.

Start with one named store, one product configuration, one fulfilment route and one date. Then build a per-unit model.

At minimum, include:

• customer sale price and discount assumption;
• marketplace referral or selling fees;
• fulfilment and storage;
• supplier price;
• freight, duty and other landed-cost inputs;
• preparation and inbound handling;
• returns and damaged-stock allowance;
• advertising or promotion assumption; and
• an explanation of what is excluded, such as fixed overhead and tax treatment.

The Amazon UK fee and revenue estimator is useful for a specified product and fulfilment method. Record the dimensions, weight, category, price, store and date used. Do not copy a remembered fee percentage into a new product model.

Use the landed-cost guide for the purchase-to-warehouse calculation. Use the Amazon Europe contribution-margin guide for the sale-price-to-contribution handoff. The wholesale profitability guide will apply this to a supplier catalogue row.

Run a downside case

At least one downside case should change more than one assumption. For example:

• the selling price falls by a realistic amount;
• fulfilment or prep costs are higher than the first estimate;
• sales are slower, extending storage and cash exposure;
• the return allowance increases; or
• advertising is needed for longer than planned.

Do not average the base and downside cases into a reassuring single number. A model is useful because it shows what would make the decision fail.

7. 6. Gate five: test the cash cycle

A product can have a positive per-unit contribution and still be a bad first order.

The reason is timing. You may pay a deposit before production, pay the balance before dispatch, wait through freight and receiving, then wait again for marketplace cash while holding enough stock to avoid a stock-out.

Record:

Link this work to the case-pack and MOQ cash-cycle guide when it is drafted. For now, keep the cash question visible in the research record. Do not leave it until after a supplier has persuaded you to buy more units.

8. 7. Gates six and seven: screen product risk and fulfilment reality

Product research is not finished when the spreadsheet turns green.

Before you request a commercial quote, screen the candidate for:

• category or product approval;
• safety and labelling obligations;
• intellectual-property and brand restrictions;
• batteries, chemicals, cosmetics, food or other regulated attributes;
• size, weight and handling limits;
• breakage, leakage or contamination risk;
• seasonality and date-sensitive stock; and
• return, disposal and customer-service exposure.

The Amazon restricted-products guide covers the platform approval question. The UK and EU product-compliance guide covers the wider product responsibility question. They are related, but they are not interchangeable.

For fulfilment, decide whether the item is realistic for FBA, merchant fulfilment or another route. A product that is profitable only if every unit arrives perfect and every return is resellable is not low risk. Use the FBA versus merchant-fulfilled guide for the operating comparison.

This stage is not a substitute for specialist legal, tax or safety advice. It is the point where you identify the question before money is committed.

9. 8. Write the decision before you contact suppliers

Create a one-page research record with these fields:

Do not write “promising” as the decision. “Request a sample after the supplier confirms the exact pack configuration” is a decision. It has an owner and a condition.

10. An illustrative example: a demand signal is not a buying decision

The following example is synthetic.

Suppose a candidate is a foldable under-sink organiser for Amazon UK. The product is defined as one unit, a stated dimension, a stated material and a stated pack configuration.

The research record shows:

• Amazon category pages suggest a visible customer need around kitchen organisation.
• A public snapshot shows several established listings.
• Google Trends shows interest that rises at some points in the year, but the comparison is relative rather than sales data.
• Reviews repeatedly mention unclear dimensions and weak hinges.
• A supplier can quote a sample, but has not yet confirmed the material specification or the carton dimensions.

The illustrative base case might look like this:

Those numbers are invented. They do not come from Amazon or a supplier.

The result is not “buy”. The contribution is thin, and the specification and carton information are unresolved. The next decision could be “request a sample and a complete quote, then rerun the model”. If the supplier cannot confirm the product identity or the downside case becomes negative, the correct decision is to reject it.

The point of the example is not the product. It is the sequence.

11. What to do next

Use the product research checklist to collect the evidence. Use the opportunity scorecard to compare candidates without hiding unknowns inside one total. Then move the surviving candidate into:

1. European supplier discovery
2. Supplier verification
3. Quotation comparison
4. Landed-cost calculation
5. Amazon contribution modelling
6. Product restrictions and compliance screening

If you do not have enough evidence to move forward, keep the candidate in research. A rejected product is not wasted work. It is a decision that protects the cash you would otherwise have committed to an untested idea.

12. Sources and review note

This article uses Amazon's public product-research and fee guidance, Amazon UK public category and Best Sellers pages, and Google's official explanation of Trends data. Amazon tools, fees and programmes can change by store, account and product. Check the current source and the signed-in Seller Central workflow before acting on a specific product.

This is an educational buying framework, not a promise of sales, approval, ranking or profit.