Guides
Revenue is the most seductive number in product research. A niche generating £500,000 per month looks far more attractive than one generating £80,000—but revenue says nothing about how much sellers keep.
A product can sell quickly and still lose money after referral fees, fulfilment charges, landed inventory cost, advertising, returns, discounts, storage and taxes. This is especially important now: a 2025 seller survey found that 67% of respondents had raised prices in response to FBA fee increases, yet almost 60% of those sellers still reported lower profits (SmartScout’s Voice of the Amazon Seller).
Recent seller accounts describe the same problem in practical terms: high sales, strong rankings and little profit once advertising and fees are reconciled.
The correct question is not “How much revenue could this product make?” It is “How much contribution profit is left under realistic—and adverse—conditions?”
Model profit per unit at the expected selling price, not the category’s headline price. Then rerun the calculation with:
- A 10–15% selling-price reduction
- Higher-than-expected advertising costs
- A realistic return and defect rate
- Storage and inbound-placement charges
- Inspection, packaging and freight costs
- Promotions and launch discounts
Amazon’s Revenue Calculator can estimate selling and fulfilment fees, but sellers must still add the costs that sit outside Amazon’s estimate.
A 30-day sales estimate can be accurate and still be dangerously misleading.
Product-research tools often display a recent monthly estimate. If that period includes Christmas, Prime Day, back-to-school demand, summer travel or a social-media spike, sellers may mistake a temporary peak for normal demand. By the time inventory has been manufactured and shipped, the buying window may already be closing.
A short data window tells you what happened recently; it does not tell you what a normal year looks like.
Review at least 12 months of search, price and sales-rank history, and preferably 24 months when data is available. Compare peak, trough and median months. Look for repeated annual peaks rather than a single unexplained surge.
Amazon explicitly recommends using Product Opportunity Explorer to identify whether products sell consistently or experience seasonal changes. External search data, such as Google Trends, can provide a second view—but marketplace purchase data should carry more weight than general web interest.
One product earning £100,000 per month does not prove that a £100,000 opportunity exists for a new entrant.
That bestseller may own most of the market because of brand recognition, patents, thousands of reviews, retail distribution, external traffic or years of ranking history. It may also be temporarily viral. Copying it means competing for the small portion of demand that remains after the dominant listing has taken its share.
One exceptional listing is an outlier, not a market.
Healthy demand is usually distributed across several relevant products. Examine how much revenue or unit volume belongs to the top one, three and ten listings. If one product captures most of the sales, the niche is concentrated and considerably riskier than its total revenue suggests.
A new seller should prefer a market where several imperfect products sell consistently—particularly where newer or lower-review listings can still win meaningful demand.
Averages hide the structure of a market.
Suppose the first ten products average £20,000 per month. That could mean every listing sells around £20,000. It could also mean one listing sells £155,000 while nine listings sell £5,000 each. The average is identical, but the opportunity is not.
The same problem applies to price, reviews, ratings and seller age. An “average of 500 reviews” might describe a reasonably balanced niche or one giant listing surrounded by lightly reviewed competitors.
The average seller does not exist. Study the distribution.
For every important metric, check:
- Median as well as mean
- Minimum and maximum
- Top-three market share
- Revenue by listing
- Review-count bands
- Price clusters
- The performance of recently launched products
Distributions reveal whether demand is broad, concentrated, polarised or accessible to newcomers.
Counting listings is not competitive analysis.
Twenty weak, interchangeable listings may present a better opportunity than five highly optimised brands. Sellers need to understand who owns the first page, why customers choose those products and what resources would be required to displace them.
Competition is not the number of rivals; it is the strength of the alternatives already satisfying the customer.
Analyse each serious competitor’s:
- Product features and materials
- Price and promotion history
- Review volume, velocity and rating
- Images, video, copy and enhanced content
- Variations and bundles
- Brand presence and off-platform traffic
- Advertising coverage
- Delivery promise
- Patents, trademarks and design protection
Pay particular attention to whether younger listings can gain traction. If every successful product is several years old and supported by thousands of reviews, “low listing count” offers little comfort.
Reading negative reviews is one of the most valuable parts of product research—but it is easy to misuse.
Not every complaint deserves a product redesign. Some are rare edge cases, delivery problems, user errors or requests that conflict with the product’s price point. Adding every requested feature can increase weight, manufacturing complexity and defect risk without improving conversion.
Amazon’s Customer Review Insights can organise recurring positive and negative themes and show how product features affect ratings (Amazon Seller Central). The analytical work, however, still belongs to the seller.
A useful product gap is frequent, important, technically solvable and commercially affordable.
Group complaints by theme, measure their frequency and compare them across several competitors. The strongest opportunities appear when the same unresolved problem affects multiple popular products.
Sales estimators are models, not audited accounts. Their accuracy varies by category, marketplace, rank stability and product type. Variations, stockouts, promotions and sudden ranking changes can distort the result.
Amazon’s own sales-estimation guidance recommends combining sales estimates with niche, search and seasonality analysis. No single estimate should carry an investment decision.
False precision creates false confidence.
Instead of forecasting that a product will sell exactly 417 units per month, create a range:
- Downside case
- Base case
- Upside case
The product should remain financially acceptable in the downside case. If it works only when the most optimistic estimate is correct, it has not been validated.
Many research models quietly assume that a new listing will capture organic sales at the same rate as established competitors. It will not.
A new product begins with weak ranking history, few reviews and limited conversion data. It may need sponsored advertising, discounts, content creation and external traffic before organic sales become meaningful. Retail media is also becoming more central to marketplace visibility, making customer acquisition part of the product’s economics rather than an optional marketing expense.
A niche is not attractive merely because demand exists; demand must be economically reachable.
Estimate likely cost per click, conversion rate and advertising cost per order. Model the launch period separately from mature operations, and calculate how much cash is required before organic ranking improves.
A product may look excellent on a research dashboard while being operationally hostile.
Large dimensions increase fulfilment and storage costs. Fragile goods create returns. Electronics introduce testing and compliance obligations. Products with many components are vulnerable to missing parts. Cosmetics, food, children’s products and batteries can carry additional regulatory requirements.
These risks are amplified when sellers commit to a large minimum order before testing actual quality and conversion. Slow stock then attracts storage costs and ties up the cash needed for advertising or replenishing stronger products.
The first order should purchase information as well as inventory.
Before placing it, confirm packaged dimensions, weight, certifications, defect tolerances, inspection requirements, lead time and realistic landed cost. Where commercially possible, begin with a smaller batch and treat the result as a controlled validation exercise.
The final mistake is psychological.
Once sellers become excited about a product, they search for evidence that confirms its potential. Strong revenue validates demand. Negative reviews validate differentiation. A rising graph validates growth. Contrary evidence is dismissed as fixable.
This is how research becomes a sales presentation rather than a risk assessment.
Good product research is an attempt to kill the idea before the market does.
Write rejection criteria before gathering data. For example:
- Minimum downside contribution margin
- Maximum top-three market concentration
- Maximum acceptable seasonality
- Evidence that newer listings can gain sales
- A recurring, solvable customer complaint
- An affordable route to first-page visibility
- No unresolved intellectual-property or compliance risk
If the product fails a critical condition, reject it. The purpose of research is not to justify launching; it is to make avoiding a bad launch cheap.
Marketplace tools are improving rapidly. Amazon says independent sellers now account for more than 60% of sales in its store, while it continues to introduce AI-assisted tools for research and operations (Amazon’s 2025 Small Business Empowerment Report). That makes data more accessible—but it also means more competitors can find the same apparent opportunities.
The advantage no longer comes from discovering a high-revenue keyword first. It comes from interpreting the market more intelligently.
Validate profit, not revenue. Measure the full seasonal cycle. Look for distributed demand. Study competitors listing by listing. Replace averages with distributions. And actively search for the reason not to launch.
A successful product is not simply one that customers want. It is one that a new seller can differentiate, acquire customers for, fulfil reliably and sell at a sustainable profit—even when the optimistic assumptions prove wrong.
10 product research mistakes new sellers keep making
# 10 Product Research Mistakes New Sellers Keep Making
After reviewing hundreds of failed product launches, the same mistakes appear time and time again. The products change, but the underlying errors remain remarkably consistent.
The latest seller discussions reinforce this pattern. New launches can generate respectable revenue while producing little or no profit once advertising, fulfilment, returns and storage are included. Meanwhile, research platforms are becoming more sophisticated: Amazon’s Product Opportunity Explorer now surfaces search, purchase, pricing, competition, review and seasonality data. Yet better tools do not automatically produce better decisions.
The main culprits: prioritising revenue over profit, ignoring seasonality in the data window, treating a single bestseller as proof of demand, skipping in-depth competitive analysis and validating opportunities with averages rather than distributions.
Here are the ten mistakes that most often turn promising product ideas into expensive lessons.
1. Prioritising revenue over profit
Revenue is the most seductive number in product research. A niche generating £500,000 per month looks far more attractive than one generating £80,000—but revenue says nothing about how much sellers keep.
A product can sell quickly and still lose money after referral fees, fulfilment charges, landed inventory cost, advertising, returns, discounts, storage and taxes. This is especially important now: a 2025 seller survey found that 67% of respondents had raised prices in response to FBA fee increases, yet almost 60% of those sellers still reported lower profits (SmartScout’s Voice of the Amazon Seller).
Recent seller accounts describe the same problem in practical terms: high sales, strong rankings and little profit once advertising and fees are reconciled.
The correct question is not “How much revenue could this product make?” It is “How much contribution profit is left under realistic—and adverse—conditions?”
Model profit per unit at the expected selling price, not the category’s headline price. Then rerun the calculation with:
- A 10–15% selling-price reduction
- Higher-than-expected advertising costs
- A realistic return and defect rate
- Storage and inbound-placement charges
- Inspection, packaging and freight costs
- Promotions and launch discounts
Amazon’s Revenue Calculator can estimate selling and fulfilment fees, but sellers must still add the costs that sit outside Amazon’s estimate.
2. Ignoring seasonality in the data window
A 30-day sales estimate can be accurate and still be dangerously misleading.
Product-research tools often display a recent monthly estimate. If that period includes Christmas, Prime Day, back-to-school demand, summer travel or a social-media spike, sellers may mistake a temporary peak for normal demand. By the time inventory has been manufactured and shipped, the buying window may already be closing.
A short data window tells you what happened recently; it does not tell you what a normal year looks like.
Review at least 12 months of search, price and sales-rank history, and preferably 24 months when data is available. Compare peak, trough and median months. Look for repeated annual peaks rather than a single unexplained surge.
Amazon explicitly recommends using Product Opportunity Explorer to identify whether products sell consistently or experience seasonal changes. External search data, such as Google Trends, can provide a second view—but marketplace purchase data should carry more weight than general web interest.
3. Treating one bestseller as proof of demand
One product earning £100,000 per month does not prove that a £100,000 opportunity exists for a new entrant.
That bestseller may own most of the market because of brand recognition, patents, thousands of reviews, retail distribution, external traffic or years of ranking history. It may also be temporarily viral. Copying it means competing for the small portion of demand that remains after the dominant listing has taken its share.
One exceptional listing is an outlier, not a market.
Healthy demand is usually distributed across several relevant products. Examine how much revenue or unit volume belongs to the top one, three and ten listings. If one product captures most of the sales, the niche is concentrated and considerably riskier than its total revenue suggests.
A new seller should prefer a market where several imperfect products sell consistently—particularly where newer or lower-review listings can still win meaningful demand.
4. Validating with averages rather than distributions
Averages hide the structure of a market.
Suppose the first ten products average £20,000 per month. That could mean every listing sells around £20,000. It could also mean one listing sells £155,000 while nine listings sell £5,000 each. The average is identical, but the opportunity is not.
The same problem applies to price, reviews, ratings and seller age. An “average of 500 reviews” might describe a reasonably balanced niche or one giant listing surrounded by lightly reviewed competitors.
The average seller does not exist. Study the distribution.
For every important metric, check:
- Median as well as mean
- Minimum and maximum
- Top-three market share
- Revenue by listing
- Review-count bands
- Price clusters
- The performance of recently launched products
Distributions reveal whether demand is broad, concentrated, polarised or accessible to newcomers.
5. Skipping in-depth competitive analysis
Counting listings is not competitive analysis.
Twenty weak, interchangeable listings may present a better opportunity than five highly optimised brands. Sellers need to understand who owns the first page, why customers choose those products and what resources would be required to displace them.
Competition is not the number of rivals; it is the strength of the alternatives already satisfying the customer.
Analyse each serious competitor’s:
- Product features and materials
- Price and promotion history
- Review volume, velocity and rating
- Images, video, copy and enhanced content
- Variations and bundles
- Brand presence and off-platform traffic
- Advertising coverage
- Delivery promise
- Patents, trademarks and design protection
Pay particular attention to whether younger listings can gain traction. If every successful product is several years old and supported by thousands of reviews, “low listing count” offers little comfort.
6. Confusing complaints with opportunities
Reading negative reviews is one of the most valuable parts of product research—but it is easy to misuse.
Not every complaint deserves a product redesign. Some are rare edge cases, delivery problems, user errors or requests that conflict with the product’s price point. Adding every requested feature can increase weight, manufacturing complexity and defect risk without improving conversion.
Amazon’s Customer Review Insights can organise recurring positive and negative themes and show how product features affect ratings (Amazon Seller Central). The analytical work, however, still belongs to the seller.
A useful product gap is frequent, important, technically solvable and commercially affordable.
Group complaints by theme, measure their frequency and compare them across several competitors. The strongest opportunities appear when the same unresolved problem affects multiple popular products.
7. Assuming estimated sales are precise facts
Sales estimators are models, not audited accounts. Their accuracy varies by category, marketplace, rank stability and product type. Variations, stockouts, promotions and sudden ranking changes can distort the result.
Amazon’s own sales-estimation guidance recommends combining sales estimates with niche, search and seasonality analysis. No single estimate should carry an investment decision.
False precision creates false confidence.
Instead of forecasting that a product will sell exactly 417 units per month, create a range:
- Downside case
- Base case
- Upside case
The product should remain financially acceptable in the downside case. If it works only when the most optimistic estimate is correct, it has not been validated.
8. Ignoring the cost of acquiring the first customers
Many research models quietly assume that a new listing will capture organic sales at the same rate as established competitors. It will not.
A new product begins with weak ranking history, few reviews and limited conversion data. It may need sponsored advertising, discounts, content creation and external traffic before organic sales become meaningful. Retail media is also becoming more central to marketplace visibility, making customer acquisition part of the product’s economics rather than an optional marketing expense.
A niche is not attractive merely because demand exists; demand must be economically reachable.
Estimate likely cost per click, conversion rate and advertising cost per order. Model the launch period separately from mature operations, and calculate how much cash is required before organic ranking improves.
9. Ordering inventory before validating operational risk
A product may look excellent on a research dashboard while being operationally hostile.
Large dimensions increase fulfilment and storage costs. Fragile goods create returns. Electronics introduce testing and compliance obligations. Products with many components are vulnerable to missing parts. Cosmetics, food, children’s products and batteries can carry additional regulatory requirements.
These risks are amplified when sellers commit to a large minimum order before testing actual quality and conversion. Slow stock then attracts storage costs and ties up the cash needed for advertising or replenishing stronger products.
The first order should purchase information as well as inventory.
Before placing it, confirm packaged dimensions, weight, certifications, defect tolerances, inspection requirements, lead time and realistic landed cost. Where commercially possible, begin with a smaller batch and treat the result as a controlled validation exercise.
10. Looking for proof instead of trying to disprove the idea
The final mistake is psychological.
Once sellers become excited about a product, they search for evidence that confirms its potential. Strong revenue validates demand. Negative reviews validate differentiation. A rising graph validates growth. Contrary evidence is dismissed as fixable.
This is how research becomes a sales presentation rather than a risk assessment.
Good product research is an attempt to kill the idea before the market does.
Write rejection criteria before gathering data. For example:
- Minimum downside contribution margin
- Maximum top-three market concentration
- Maximum acceptable seasonality
- Evidence that newer listings can gain sales
- A recurring, solvable customer complaint
- An affordable route to first-page visibility
- No unresolved intellectual-property or compliance risk
If the product fails a critical condition, reject it. The purpose of research is not to justify launching; it is to make avoiding a bad launch cheap.
The better product-research standard
Marketplace tools are improving rapidly. Amazon says independent sellers now account for more than 60% of sales in its store, while it continues to introduce AI-assisted tools for research and operations (Amazon’s 2025 Small Business Empowerment Report). That makes data more accessible—but it also means more competitors can find the same apparent opportunities.
The advantage no longer comes from discovering a high-revenue keyword first. It comes from interpreting the market more intelligently.
Validate profit, not revenue. Measure the full seasonal cycle. Look for distributed demand. Study competitors listing by listing. Replace averages with distributions. And actively search for the reason not to launch.
A successful product is not simply one that customers want. It is one that a new seller can differentiate, acquire customers for, fulfil reliably and sell at a sustainable profit—even when the optimistic assumptions prove wrong.