Skip to content

Finding Keywords

Pull candidate keywords from several independent sources before you judge any of them.

Cast the net wide

Keyword research separates gathering from judging, and collapsing the steps is the most common mistake. First you gather — widely, uncritically, from every source that shows you real shopper language. Only later do you filter. Judging a term the moment you find it discards the ones that look weak alone but earn their place once you see the full set.

The point of using independent sources is that each is blind to what the others surface. Autocomplete shows phrasing; competitors show positioning; search-frequency data shows demand; your own reports show what actually converts. A term that appears in only one of them is not automatically weak — it may be the gap nobody else is covering.

Gather first, judge later
A keyword that looks irrelevant beside one source often makes sense beside another, so collect everything before you cut anything.

Pull from each of the following, and keep the raw list intact:

Autocomplete

the words shoppers type before they finish typing.

Competitor listings

the terms rivals have chosen to rank on.

Search-frequency data

aggregate demand, ranked by how often a phrase is searched.

Advertising reports

the queries that already spent money on your product.

Adjacent phrasing

synonyms and substitutions a shopper might reach for instead.

Marketplace autocomplete

Start typing a seed term into the marketplace search bar and read what it suggests. These completions are drawn from real queries, so they reflect how shoppers phrase intent rather than how you would describe the product. Work through several seeds — the category, the problem it solves, the material, the use case — and record every suggestion, including the ones that seem off. Autocomplete rewards breadth: one unexpected modifier can open a whole cluster of terms.

Competitor listings

The listings ranking for your seed terms are another independent read on the same market. Study the titles, bullets, and descriptions of the products you would compete against, and note which terms recur across several of them. A phrase that several established sellers all target is one the market has already validated. Terms that appear on only one listing are worth keeping too — they may be that seller's edge, or their blind spot.

Brand analytics search frequency

The marketplace's own brand analytics expose aggregate search-frequency rankings — the closest thing to a demand signal you will get directly from the source. This data sits behind Brand Registry, which opens once the marketplace accepts your trademark filing; the earlier guide on Trademark and Brand Registry covers how to reach that gate. Until you have access, lean on the other sources, and treat this one as the read you unlock later.

A keyword you cannot fulfill is not a keyword

Demand you cannot serve — wrong variation, restricted category, thin margin — belongs on a separate list, never your target list.

Your own search-term reports

Once campaigns are running, the advertising search-term report becomes the most honest source you have: the actual queries that triggered your ads, and which of them converted. It only exists after enough operating history to have spent on real traffic, so it arrives last — but it eventually outranks the others, because it reflects behavior on your listing rather than the market at large. Feed its winners back in and let it correct your earlier guesses.

Working the sources

Run the same loop each round so the list stays comparable:

1

Collect

Pull from every source available to you now, keeping seeds and their suggestions in one place without editing.

2

Merge

Combine the passes into a single list and remove exact duplicates, preserving near-duplicates — small wording differences can rank separately.

3

Flag

Mark any term you cannot fulfill on, so it never reaches the stage where you score and prioritize.

Cast wide here; the judgment comes next.

Was this page helpful?