The usual advice is to buy a keyword tool and export a list. For a store with a handful of products and no rankings, that produces a spreadsheet of numbers you cannot act on, most of them for queries you have no chance of reaching.
There is a better first pass, it is free, and the raw material is data you already own.
Start where the data already is
Four sources, all of them yours, all of them more specific to your business than any tool.
1. Your own order history
Every product you have sold, and how many. This tells you which products deserve their own real page and which are marginal.
Look for the pattern rather than the totals. Which category sells most? Which variation? Which price band? A store's structure should reflect what people actually buy from you, not what you hoped they would.
2. Your buyer messages
This is the most underused asset in the whole subject. Every question a buyer has ever sent you is a query somebody typed somewhere.
Read the last hundred messages and write down every question. Sizing, materials, care, gifting, timing, compatibility, personalization. That list is the content plan for your store, in the buyers' own words, and no tool will generate it because it is specific to you.
3. Your marketplace search terms report
Most platforms show sellers, in their own stats, the search terms that led to their listings. That is a record of the actual language of people who bought from you.
It is not directly transferable to open-web queries and it is excellent vocabulary. It tells you which words your buyers use for the thing you sell, which is the hardest part of this to guess.
4. The search box you are researching in
Type a phrase and read the suggestions. Search it and read the related searches at the bottom, and any question boxes on the way down. This is free, it reflects real queries, and it is more current than most datasets.
Then look at who ranks. That second step is the one people skip and it is where the useful information is.
Sort everything into four buckets
This is the step that turns a list into a plan, and it is the one a tool export cannot do for you.
Product queries. Someone describing a specific thing they want. These map to product pages.
Category queries. Someone describing a type of thing, still choosing. These map to collection pages, and they are usually where the volume is.
Question queries. Someone who needs to know something before they can buy, or after. These map to content, and they are frequently the only queries a new store can realistically reach early.
Brand queries. Yours, or someone else's. Yours are the ones to watch over the next year as a measure of whether anything is working.
A query in the wrong bucket produces a page that cannot win. A product page aimed at a question query loses to an article; an article aimed at a product query loses to a shop.
The winnability check, which replaces the difficulty score
Tools sell a difficulty number. You can do better with your own eyes, and it costs nothing.
Search the query and look at the first page.
Large national retailers and marketplaces occupying every slot? Not reachable in the near term. Note it and move on; there is no shame in a query being out of range and no benefit in aiming at it.
A mix that includes small independent sites, blogs and forum threads? Reachable. This is your target set.
Pages that are thin, old, or clearly not answering the query well? That is an opening, and it is the best signal available to a small site.
Nothing that looks like your kind of page at all? The engine has decided this query wants something else, and you should believe it.
Doing this by hand for thirty queries takes an afternoon and produces a better target list than a tool's export, because it accounts for what the results actually look like rather than reducing them to a number.
Go long, deliberately
The single most useful habit for a small store is to aim at longer, more specific queries.
A broad category term is contested by everyone. A specific one, with a material, a use, an occasion or a recipient attached, is contested by far fewer people and is typed by a person much closer to buying.
You are unusually well placed to find these, because your listings already taught you the modifiers your buyers use. That vocabulary is the asset a general keyword tool does not have.
And a page aimed at a specific query is easier to write well, which matters more than it sounds, because the pages that win are the ones that answer the query better rather than the ones that mention it more.
The pages this produces
Run the process and you will end up with a build list rather than a spreadsheet.
A handful of collection pages covering your main category terms, each with real text on it rather than a bare grid.
Product pages that are actually written, rather than listing copy transplanted. What has to change is covered in our guide to turning a marketplace listing description into a product page.
A set of question pages taken directly from your own buyer messages.
And an ordering, because you cannot build all of it at once. Start with the collection pages, since they cover the most ground per page.
What this method does not do
It does not give you search volume, and for a new store that is a smaller loss than it seems. A query with modest volume that you can actually reach is worth more than a large one you cannot, and the winnability check above is the part that decides which is which.
It also does not tell you how long anything will take, and this site does not publish a timeline for that as its own prediction. The variables that decide it are set out in our article on what actually drives the wait for organic traffic.
When a tool becomes worth paying for is once you have rankings to track and a catalogue large enough that manual checking stops scaling. Before that, the constraint is not data. It is pages.
For where this sits in the wider decision, our pillar on whether you should move at all is the place to start.