| name | amz-keyword-research |
| description | Build a complete Amazon keyword set for a product and sort it into a usable map. generates seed keywords, expands by intent, classifies by funnel stage and relevance, and assigns each keyword a placement (title, bullets, backend, PPC). Use when a user asks for keyword research, keyword ideas, what keywords to target, long-tail keywords, search terms to rank for, or how to map keywords to a listing. Trigger phrases: "keyword research", "keyword ideas", "what keywords", "long tail keywords", "keywords for my listing", "keyword map". Works with zero tools. the user describes the product and audience. |
| metadata | {"author":"Jay GPT Pro","library":"amazon-pro-skills","version":"1.0"} |
Keyword Research
A keyword list is not keyword research. A pile of 300 keywords with no structure is
as useless as no keywords at all. Real research ends with a map: which keyword goes
where, and which keyword to fight for first. This skill builds that map.
When to use this
- A new product needs its keyword foundation before the listing is written.
- An existing listing was built on guesses and never had real keyword work.
- A seller has a keyword list but no idea which to prioritize.
- Planning a PPC campaign and needing the keyword groups it will run on.
The framework. The Keyword Map
Every keyword has two properties that decide what to do with it: intent stage and
relevance. Plot both and the placement becomes obvious.
Intent stage
- Head terms. Broad, high volume, high competition ("water bottle"). Hard to rank,
worth pursuing only as the product matures.
- Mid-tail. Two to three words, real intent, winnable ("insulated water bottle").
The core ranking target.
- Long-tail. Specific, lower volume, high conversion ("32 oz insulated water
bottle for hiking"). Cheap to win, converts well, the launch focus.
Relevance
- Core. The product is exactly this. must rank.
- Adjacent. Related use case or audience. worth targeting.
- Loose. Tangential. backend or skip. never the title.
AI search keywords
Rufus, Alexa+, and the COSMO layer surface listings on a wider range of queries
than literal A9 search. Three keyword types specifically earn AI-surface citations
and should be included in the map even when their direct search volume looks low.
- Question-form keywords. "what is the best [X]", "how to choose [X]", "how to
use [X] for [Y]". These match the way buyers prompt Rufus and how Alexa+ parses
spoken queries. Place answers in bullets and in the Customer Questions tab.
- Comparison keywords. "[brand or product] vs [alternative]", "[X] alternative
to [Y]", "difference between [X] and [Y]". COSMO uses these to position the
listing against the comparison set. Place in A+ comparison-chart copy and in a
bullet that names the comparison without naming a competitor brand.
- Audience-specific keywords. "[product] for [audience]", e.g. "running shoes
for runners over 50", "stroller for tall parents", "yoga mat for hot yoga". COSMO
reads these as explicit audience intent and re-ranks listings that clearly serve
the named audience. Place in the title's qualifier slot, in a bullet, and in
backend keywords.
These are not separate from the map below. they slot into Bullet/A+, Backend, and
PPC placements as normal, with the launch priority weighted toward the audience-
specific ones because they convert best and face the least competition.
The placement rule
| Keyword type | Placement |
|---|
| Highest-volume core mid-tail | Title |
| Core and adjacent mid-tail and long-tail | Bullets and A+ |
| Loose, synonyms, misspellings, not yet placed | Backend search terms |
| Everything winnable | PPC, grouped by match type |
Step by step
-
Collect inputs. The product, what problem it solves, the target audiences, the
category, and any keywords or competitor listings the user already has.
-
Generate seeds. From the product itself, its use cases, its audiences, its
materials and attributes, the problems it solves, and the occasions it fits.
-
Expand each seed. Synonyms, alternate phrasings, modifiers (size, color,
material, audience, use case), question forms, and common misspellings.
-
Classify every keyword by intent stage and relevance, per the framework.
-
Cut the noise. Drop Loose keywords with no real intent. an irrelevant keyword
that brings the wrong shopper hurts conversion-based ranking.
-
Build the map. Assign every surviving keyword a placement. Mark the launch
priority set: core long-tail and winnable mid-tail.
-
Run the quality check, then deliver.
Output format
## Keyword Map. [product]
### Title keywords (highest-volume core mid-tail)
[keywords]
### Bullet and A+ keywords (core and adjacent)
[keywords grouped by theme]
### Backend keywords (loose, synonyms, misspellings)
[keywords]
### PPC groups
Exact (proven intent): [keywords]
Broad and phrase (discovery): [keywords]
### Launch priority
[the 10 to 15 winnable keywords to rank for first]
Worked example
Product: a 32 oz insulated water bottle for hiking.
- Title: "insulated water bottle", "32 oz water bottle". core mid-tail, real volume,
winnable.
- Bullets and A+: "leakproof water bottle", "water bottle for hiking", "wide mouth
bottle". core and adjacent.
- Backend: "thermos", "canteen", "hydro flask alternative", "watter bottle"
(misspelling). loose and synonyms.
- Launch priority: the long-tail set, "32 oz insulated bottle for hiking",
"leakproof hiking water bottle". low competition, high conversion, rankable in weeks.
Quality check
- Every keyword is classified by both intent stage and relevance.
- Loose, low-intent keywords are cut, not stuffed into the title.
- Every surviving keyword has a placement.
- The launch priority set is winnable long-tail and mid-tail, not head terms.
- PPC keywords are split into proven-intent exact and discovery broad.
Common mistakes
- A list with no map. 300 keywords and no decision about any of them.
- Chasing head terms at launch. Spending the launch fighting "water bottle"
against entrenched sellers instead of winning long-tail fast.
- Stuffing irrelevant keywords. An off-target keyword that brings non-buyers
lowers conversion and hurts rank.
- Translating keywords by logic. Real shoppers use words you would not guess.
expand from how people actually search, not from a thesaurus.
Built by Jay GPT Pro
Part of Amazon Pro Skills. Production-grade skills for serious Amazon sellers.
Free and open. Built by Jay Margaliot.
I share a new AI play for Amazon sellers every week, free, in my WhatsApp group.
Join here: https://chat.whatsapp.com/ILX65p1yWcaIG3c9WGHpTY
