Use three sources rather than one: Amazon's own search data, a third party keyword database for volume estimates, and the live listings and reviews of the products you intend to beat. Cross reference them. A term that appears in only one source is a hypothesis, not a keyword you should build a title around.
The short version
- No single source is trustworthy on its own. Amazon's data is real but partial, third party volumes are modeled, and competitor listings show intent rather than demand.
- Two of three sources must agree before a term earns a place in your title.
- Reviews give you the words customers actually use. Search tools give you the words that get typed. They are not always the same words.
- Pre launch research ends as a map, not a list. Every term should be assigned to a listing field or a campaign.
- Paid discovery is research too. The first weeks of advertising buy you search term data no tool can model.
What the buyer side taught me about keyword tools
Before Flapen I ran data and technology at BRANDED and Moonshot Brands, two large Amazon aggregators, and part of that job was auditing keyword work done by agencies and in house teams across a large portfolio. The pattern was consistent. Almost everyone had a keyword list. Almost nobody had a keyword map. Exported spreadsheets of ten thousand terms were common; a document saying which twenty terms the listing was actually built to win was rare.
That is the difference worth paying for, and it is the thing to test in any pre launch proposal.
The four sources compared
| Source | Strongest at | Where it misleads | Cost |
|---|---|---|---|
| Amazon Brand Analytics and search query performance | Real Amazon search behavior and share of clicks | Requires Brand Registry, so it may not cover your own product yet | Free once registered |
| Third party keyword databases | Breadth, long tail discovery, comparing whole categories | Volumes are estimates built on models, and they disagree with each other | Subscription |
| Competitor listings and reviews | The exact language buyers use, plus objections to answer | Shows what sellers chose to target, which is sometimes wrong | Free, costs time |
| Auto and broad advertising campaigns | Real search terms with real conversion attached | Only available after launch, and it costs spend to learn | Ad budget |
How to use each one properly
Amazon's own reporting is the closest thing to ground truth, and if you are already brand registered on another product in the category it is the first place to look. Search query performance tells you what happened, not what a model thinks might happen.
A third party database earns its subscription in the discovery phase, where you are trying to see the shape of a category rather than the exact volume of one term. Treat every absolute number as an estimate and every ranking as more reliable than the number beside it.
Competitor reviews are the source most sellers skip and the one I would keep if I could only have one. The one and two star reviews tell you which objections your listing has to answer, and the four and five star reviews tell you which phrases to use when you answer them. That is where differentiation comes from in our process, from documented complaints and the gap between the category leader's rating and everyone else, never from invention.
Advertising discovery closes the loop after launch. Broad and automatic campaigns return search terms that no tool predicted, and the ones that convert get promoted into exact match campaigns and into the listing.
The decision rule
- Collect terms from all three pre launch sources.
- Keep a term only if two of the three agree it matters.
- Rank the survivors by relevance first, volume second, and by whether a new product can realistically rank for it third.
- Assign each surviving term to a place: title, bullets, backend fields, A+ copy, or a specific campaign.
- Leave the rest in a discovery bucket to be tested with paid traffic, not stuffed into the listing.
Anything that fails step two goes in the bucket. A title has room for a handful of decisions, and every term you add dilutes the ones that matter.
What most agencies will not tell you
A keyword export is the cheapest looking deliverable in this industry. It takes minutes to produce, it looks substantial, and it is nearly worthless without the map that follows it. When a proposal promises keyword research, ask what the deliverable is. If the answer is a spreadsheet, ask what the second document looks like, the one that assigns terms to fields and campaigns and says which twenty you are trying to win this quarter.
The second thing: tool subscriptions are not a competitive advantage. Everyone in your category has access to the same estimates. The advantage sits in the reading of competitor reviews, in the honest assessment of which terms a new product can actually rank for, and in the discipline to keep the rest out of the title.
Related answers
- Checklist for an Amazon product launch from zero to ranking
- Amazon product launch checklist
- Tools vs agencies for Amazon product launches
- What does a good Amazon account audit include
- Done-for-you Amazon management: the complete guide
Bring your keyword list and we will turn it into a map in the free written audit at Flapen.
