Keyword research for a store is not a list of phrases. It is a map that connects every phrase and every full question a buyer might use to the one page on your site that should answer it.
A list of two thousand keywords sorted by search volume is worth almost nothing on its own. It tells you what people search for. It does not tell you which page should target which phrase, which phrases you already rank for and are wasting, or which questions your competitors have answered and you have not.
The map is the deliverable. Everything else is a spreadsheet.
Why Do Stores Need Different Research From Other Businesses?
Because a store has two page types that must own two different kinds of search, and mixing them up wastes both.
Collection pages should own the broad, plural, comparing kind of search. Somebody looking for a category of thing, still deciding, wanting to see options. These searches carry the most volume and the most future revenue.
Product pages should own the narrow, specific, decided kind of search. A model name, a size, an exact specification. Lower volume, much higher intent, and often much easier to win.
Stores get this wrong constantly. Product pages are optimised for broad category terms they will never win, and collection pages are left with a headline and nothing else. The result is a catalogue competing against itself while ranking for neither.
Getting the split right is most of the value in this work, and it costs nothing extra to do correctly the first time.
How Do You Research Questions Rather Than Keywords?
We collect the full sentences buyers use, from your own site search, your support inbox, product reviews, community forums, and from the AI tools themselves.
This is where prompt research differs from keyword research. A keyword tool reports what was typed into a search box. It has almost no visibility into what people type into a chat tool, which is longer, more conversational and often more revealing about what the buyer actually needs.
- Your internal site search. The single most underused source in ecommerce. It tells you exactly what visitors expected to find and in their own words.
- Your support inbox and live chat. Every repeated pre purchase question is a page you have not written yet.
- Product reviews, yours and your competitors. Reviews explain why people bought, what they compared, and what nearly stopped them.
- Forums and communities where your category is discussed, where people ask questions they would be embarrassed to ask a salesperson.
- The AI tools themselves. We ask them the buying questions in your category and read what they say, which shows both the phrasing people use and the answers currently being given.
What Does the Finished Map Look Like?
One row per target page, with the main phrase it should own, the supporting phrases, the full questions it must answer, and its current position.
Every page on your site gets a row. Pages with no phrase worth owning get flagged for merging or removal, because a catalogue full of pages targeting nothing is a drag on everything else.
| Page | Main phrase | Questions the page must answer | Now |
|---|---|---|---|
| Collection page | The broad category term buyers compare within | What are the types, how do I choose, what should I check, what does it cost | Position 14, no copy on page |
| Buying guide | The how do I choose version of the same subject | The full comparison, with the collection page linked from it | Does not exist |
| Product page | The exact model or specification name | Will it fit, how long does it last, what is included, how does it compare to the obvious alternative | Position 6, manufacturer copy |
| Answer page | The definition or comparison question | One question, answered completely, in a format a machine can lift | Does not exist |
How Often Should Research Be Redone?
A full rebuild once a year, a refresh every quarter, and a small check every month for new questions appearing in your category.
The core of a category does not change quickly. What people call the products they buy is fairly stable, and a map built well will hold most of its value for a year.
What changes faster is the edge. New models, new competitors, new problems, new phrasings, and now new questions being asked of AI tools that did not exist a year ago. Those appear continuously and the ones that matter should be picked up within weeks, not at the next annual review.
So the monthly check is narrow. What is new, what is rising, and is anything we already targeted no longer worth targeting. It takes little time and it stops the plan quietly going stale.
Can You Use the Research Without Us?
Yes. The map is a working document written for whoever will act on it, whether that is your team, a freelancer or another agency.
Each row tells you the page, the phrase, the questions and what to do next. There is no proprietary format, no tool you have to keep paying for, and nothing hidden behind our login.
Plenty of stores buy the research once, hand it to their own writer and work through it over a year. That is a legitimate way to use it, and for a store with a good writer already on staff it is often the right one.