Before launch you have no data of your own, so use three substitutes: competitor term
extraction from a research tool, Amazon autocomplete for real phrasing, and competitor
reviews for vocabulary. Treat the whole list as a hypothesis you will replace within 90 days
of selling.
The short version
- You have no data yet. Everything pre-launch is an estimate.
- Three sources: competitor terms, autocomplete, competitor reviews.
- Group into rank-for, cover, and buy-only. A flat list is not a strategy.
- Plan to replace it with your own search term data after 90 days.
- Do not over-invest in pre-launch keyword work. It has a short shelf life.
The three pre-launch sources
I run Flapen with 50 operators managing about 70 brands. Pre-launch keyword work is
harder than post-launch, because the best source does not exist yet.
| Source | What it gives you | Limitation |
|---|---|---|
| Competitor term extraction | A broad candidate list quickly | Estimated volume, no intent signal |
| Amazon autocomplete | Real phrasing buyers type | No volume data |
| Competitor reviews | Vocabulary and unmet needs | No volume data |
| Brand Analytics | Real search frequency rank | Needs Registry, limited pre-launch use |
Autocomplete is the most underrated of the three and it costs nothing. Type your category into
the Amazon search bar and watch what completes. Those are live, real phrasings, unfiltered by
anyone's volume model.
Why the pre-launch list is a hypothesis
Estimated search volume is a model of what buyers might type. Your search term report, once
you are selling, is a record of what they did type and then buy.
The gap between those two is routinely large. Terms that look excellent by volume convert
poorly for your specific product. Terms nobody predicted turn out to carry your best
conversion rate. Neither outcome is knowable in advance.
Which is why pre-launch keyword work should be good enough rather than exhaustive. A
spreadsheet of 2,000 terms with volume estimates takes days to build and gets replaced within
a quarter. Spend that time on the primary image instead, which caps every click your listing
will ever receive.
Grouping matters more than list length
Split candidates into three sets before writing anything.
Rank-for. A small number of terms you can realistically achieve organic position on. These
drive the title and the first bullet.
Cover. Terms worth being indexed for but unlikely to rank on early. These go in bullets,
description, and backend search terms.
Buy-only. Terms you will reach through advertising and probably never rank on organically,
usually the broadest and most competitive.
Most beginners produce a flat list and then stuff the title, which reads badly and converts
worse. The grouping is the strategy.
Reading competitor reviews for keywords
Underused because it looks like qualitative work rather than keyword research.
Read fifty negative reviews on the top competing products. You are looking for two things: the
words buyers use for the product and its features, which often differ from industry
vocabulary, and the recurring complaints, which are the objections your listing must
pre-empt and frequently contain search-relevant phrasing.
This also feeds differentiation. The market states where it wants something better, and
innovating only where it is asking is more reliable than invention.
What to do at 90 days
Replace the whole thing.
Pull your search term report, sort by conversion rate rather than by spend, and rebuild the
title, bullets, and backend terms around what actually converted. Most sellers never do this,
which is why a listing built on pre-launch estimates is still running on them a year later.
Set the calendar reminder now, while you are doing the pre-launch work, because nobody
remembers to revisit a job they consider finished.
What most agencies will not tell you
Pre-launch keyword research is one of the easiest deliverables to make look substantial. A
long spreadsheet with volume columns and color coding takes little time and is largely
unusable, because the hard part is prioritization rather than collection.
Ask a provider which few terms they think you can realistically rank for, and why those. That
judgment is the deliverable. The list is raw material.
The second thing: the shelf life is short and nobody mentions it, because a research
deliverable that expires in ninety days is harder to sell than one presented as a foundation.
Ask explicitly what happens to the keyword strategy once your own data exists.
Related answers
- What products to use for Amazon keyword research
- What to use for listing optimization on Amazon
- How to launch first product on Amazon
- Amazon listing optimization agency checklist
- Hiring an Amazon agency: the complete guide
Set the 90-day reminder to rebuild from your own data. We do it for clients at Flapen.
