Five manual methods work: mining one-star and two-star reviews in categories you know, scanning for markets where every leader sits below 4.5 stars, reading niche forums and subreddits for repeated complaints, walking retail shelves for products missing on Amazon, and logging your own purchase frustrations. Generate twenty ideas and expect most to die under scrutiny.
The short version
- You do not need software to find ideas. You need a reading habit and a tally sheet.
- Complaints are the raw material. Every method below is a different place where complaints accumulate.
- Twenty in, one or two out. High mortality is the sign the process works, not that it failed.
- Ideas are cheap. Discipline is the asset. Attach a kill condition before an idea touches money.
- Stay where you have judgment. Categories you understand let you tell a fatal complaint from noise.
Where ideas actually come from
You want to build on Amazon and you do not have a product idea. That is a better starting position than the seller with one idea he is in love with, because you can still be ruthless. Products worth launching are not invented. They are found where buyer frustration has already pooled, and all five methods below are just different pools.
The five methods
- Mine one and two star reviews in categories you know. Done properly means a written tally of complaints grouped by theme across the top listings, not a skim. The recurring, fixable complaint is a product spec another factory run can solve.
- Scan for rating gaps. Walk a category's best sellers and note the markets where nobody clears 4.5 stars. Done properly means also asking why: a market where physics is the problem is a warning, one where laziness is the problem is an invitation.
- Read the forums where the hobby lives. Subreddits, groups, specialist boards. Done properly means collecting complaints that repeat across threads months apart, because a one-thread gripe is weather and a twelve-month gripe is climate.
- Walk physical retail. Supermarkets, hardware stores, specialist shops. Done properly means noting what people buy offline that is underserved, clumsily listed, or badly bundled online, then checking the Amazon shelf against it.
- Log your own frustrations for a quarter. Every product that disappoints you goes into a note with the specific failure. Done properly means the note records the failure, not the product, because the failure is the idea.
No idea survives without a kill condition
Generation is the easy half, which is exactly why it gets romanticized. Before any idea on your list costs money, write its kill condition: the rating trend, return rate, conversion rate, or acquisition cost trajectory that, inside a defined window, sends it to the bin. This is our scale, fix, or kill discipline, and it exists because I once spent three months feeding advertising money into a failing product, waiting for a turnaround that never came. The criteria are the receipt from that mistake. An idea plus a kill condition is a plan. An idea alone is a hope with a budget. When a shortlist clears this bar, the structured way forward is a small, capped first order, which is how our Amazon FBA launch program takes every new idea to market.
What most agencies will not tell you
Product picking gets romanticized because research retainers sell well, and mystique justifies the invoice. The truth is less flattering to everyone: idea generation is nearly free, the methods above cost attention rather than money, and the scarce skill is elimination. A firm that hands you ten "validated winners" and no kill conditions has sold you lottery tickets with a cover page. Ask any research partner what evidence would make them tell you to drop their own recommendation. The good ones answer instantly, because they wrote it down before they pitched.
Related answers
- AI tools to generate Amazon product ideas
- Competitor analysis workflow for new Amazon sellers
- Validate product ideas before launching on Amazon
- Step-by-step Amazon launch with zero product idea
- Amazon seller roadmaps and capital: the complete guide
When your shortlist is ready for kill conditions and a launch plan, bring it to Flapen.
