Tools answer three questions in order: is the market big enough, is the demand durable, and is there a rating gap you can win. A research suite plus Amazon's own data covers the first two. The third comes from reading competitor negative reviews by hand, which no software does for you.
The short version
- Sequence beats software. The order you ask the questions matters more than which subscription you hold.
- Estimates are for ranking candidates, not for forecasting revenue. Treat every figure as relative.
- Durability is the question people skip. A category that spiked last quarter is not a category.
- The decisive input is unstructured. Complaints in one-star reviews are where differentiation comes from.
- Free sources carry more weight than sellers expect. Amazon's own pages are the most reliable data you have.
The research sequence, with a gate at each stage
Each stage kills candidates. That is the point. About nine in ten ideas should not survive to stage five.
- Generate candidates. A research suite with category filters, plus your own browsing of best seller lists and new release pages. Gate: a written list of candidates with the category and price band recorded.
- Size the market. Estimated revenue across the top listings for the main terms, cross-checked against category depth. Gate: an annual figure you would be willing to defend to somebody else.
- Test durability. Look at a demand trend across at least a year using seasonality and trend data. Gate: the demand exists outside a single season or a single spike.
- Read the competition properly. Review counts, rating distribution, listing quality and how long the leaders have held position. Gate: at least one leading product with a visible weakness rather than five with none.
- Find the rating gap by hand. Read the one and two-star reviews on the top listings and cluster the complaints. Gate: a specific, repeated complaint you can design against and photograph.
- Check the economics. Amazon's own fee calculator with real dimensions and weight, plus landed cost and an advertising allowance. Gate: contribution margin that survives a realistic acquisition cost.
- Check the constraints. Category approval, hazardous materials, patents, compliance and intellectual property risk. Gate: nothing that stops you shipping.
Skipping stage three and stage five is how sellers end up in a crowded category with a product identical to four others and no reason for anyone to choose it.
Which category of tool serves which stage
I will not rate named products, because I cannot verify anyone's current data accuracy in your category and the situation changes constantly. Judge them by stage instead.
| Stage | Tool category | What to be skeptical about |
|---|---|---|
| Candidate generation | Research suites with filters and databases | Filters produce the same shortlists for everybody using them |
| Market sizing | Revenue estimation on top listings | Estimates are inferred, so use them comparatively |
| Durability | Trend and seasonality data, including general search trend tools | Short histories hide seasonality entirely |
| Competitor analysis | Listing and review analysis, reverse lookups | Review count is a lagging indicator of strength |
| Rating gap | Your own reading, optionally clustered by text analysis | Summaries drop the specific detail that matters |
| Economics | Amazon's fee calculator with real measurements | Dimensional weight and packaging change the answer |
| Constraints | Seller Central category requirements, patent databases | Absence of a warning is not the same as clearance |
The part that decides everything
Our sourcing frameworks were built across more than 500 brands, and the strongest pattern in that history is unglamorous: the products that worked were differentiated from what competitors were being criticized for, never from what someone thought would be clever. Invention is expensive and usually wrong. Answering a complaint that thirty existing customers wrote down in public is cheap and usually right.
That is why stage five cannot be automated away. A text summarizer will tell you customers mention durability. Reading the reviews tells you the handle snaps at the join after about four months, which is a design instruction, a photograph, and a bullet point on your listing.
What most agencies will not tell you
The polished research report is the easiest deliverable to fake. Screenshots of estimated revenue, a competitor table and a recommendation can be assembled in an afternoon by someone who has never sold in the category.
The way to test it is to ask what would have disqualified the recommendation. A real process has kill conditions at every stage, and someone who ran one can tell you which candidates died at stage three and why. We look at 90 or more data points before committing to a product, and the useful half of that work is not the data collection, it is the willingness to discard a category everyone is excited about because the durability check failed.
Related answers
- Keyword research tools for Amazon private label
- How to validate a product before Amazon launch
- What products to start with for first Amazon private label
- Recommended software for Amazon brand rollout
- Amazon launch services: the complete guide
If you want a second opinion on a candidate product before you order it, ask at Flapen.
