Score every replacement on one axis: does it report first party data or modeled estimates. Your advertising search term report, Brand Analytics search query data, the ad console's own suggestions, and the live search results page are all first party. Rank candidates on that before you compare features or price.
The short version
- First party beats modeled, always. Your own search term report is the only keyword source that knows what actually converted for you.
- A keyword list is not a plan. Without a target ACoS attached to each stage, the list just tells you where to lose money faster.
- Free sources cover the core job. Search terms, Brand Analytics, autocomplete and the results page answer most questions a subscription answers.
- Paid tools buy speed and breadth, not truth. Worth it when you screen many categories. Rarely worth it for one listing.
- Language matters more than volume. In German, Spanish or French, translated keyword lists rank badly no matter which tool produced them.
Score the alternatives before you pick one
I am not going to grade any named vendor's accuracy. I cannot verify their numbers for your category, and nobody selling you a comparison can either. What I can give you is the scorecard we use internally, which works on any source.
| Criterion | Weight | What a 5 out of 5 looks like |
|---|---|---|
| Source of truth | 30% | First party data from Amazon or your own account, not an estimate |
| Action it produces | 25% | Tells you what to bid or what to write, not only what exists |
| Refresh rate | 15% | Updates at least weekly so seasonality is visible |
| Marketplace and language coverage | 15% | Native data in the marketplace you actually sell in |
| Cost per decision | 15% | Cheap relative to the decision it informs |
Score each candidate out of five per row, multiply by the weight, and total it. Anything under about 3.5 is a research toy rather than a research tool. Run your current subscription through the same grid before you renew it.
The first party sources, scored highest for a reason
Search term reports from your advertising console. Every dollar you have spent has already bought you a list of the exact phrases shoppers typed, with impressions, clicks, spend and orders attached. This is the highest scoring keyword source that exists for a live account, and most sellers never open it.
Brand Analytics search query data. Available once you are brand registered. It shows what buyers search inside your category, how the clicks distributed, and where your share sits against the rest of the page. It answers the question a volume estimator only guesses at.
The search results page itself. Type the phrase and read what Amazon returns. If the top results are a different product than yours, the keyword is not yours no matter how large the estimated volume is. This takes minutes and it is free.
Autocomplete and related searches. The suggestions Amazon offers as you type are drawn from real queries. They are a fast way to find the modifiers buyers attach to your root term.
Your own competitors' listings. Read the titles and bullets of the products that sit above you. Where the same phrase appears across several strong listings, it is doing work.
The paid research route, honestly
A subscription is a speed purchase. It compresses a week of manual work into an afternoon and covers categories you do not sell in yet, which is exactly what you need when you are screening. If you have one product and one category, the same money buys a small exploratory ad campaign that returns real search terms tied to your own conversion rate rather than a modeled one.
Why the ACoS target decides the keyword list
Here is the part the tooling conversation usually skips. A keyword is not good or bad on its own. It is good or bad relative to what you are trying to do with the product right now.
At launch, the job is to buy visibility and rank, so the ACoS target is deliberately aggressive and you will bid on head terms you cannot yet win profitably. At maturity, the job is efficiency, and many of those same head terms should be cut in favor of long-tail phrases that convert. The keyword did not change. The stage did.
So before the list, write down the stage and the target. Then sort the list into three buckets: terms that build rank, terms that make money, and terms that do neither and should be negatives. Any tool that hands you volume without asking which bucket you are filling has given you a spreadsheet, not a decision.
We keep an in-house technology team building our own advertising and marketing tools for exactly this reason, and even with that the ranking work is done by an operator reading the search term report by hand every week.
What most agencies will not tell you
The software line item is often the easiest part of the invoice to justify and the least connected to results. An agency can show you an impressive keyword export in a pitch, and that export costs them almost nothing to produce. The work that moves rank is negative keyword pruning, match type discipline, and rewriting the listing so the keyword you are paying for actually converts once the shopper arrives.
Ask two things. Ask for the launch ACoS number and the maturity ACoS number for a product like yours, and check that they are different. Ask whether they read your search term report weekly or run monthly bulk uploads. The first question tells you whether they think in stages. The second tells you whether anyone is actually looking at your account.
Related answers
- How to rank new Amazon listings from scratch
- Typical Amazon PPC budget for new listings
- Alternatives to Jungle Scout for product research
- Optimize Amazon listings for Q4
- Amazon seller roadmaps and capital: the complete guide
We set ACoS targets by product stage, and you can ask us for yours at Flapen.
