How to Do Keyword Research for an Ecommerce Website
A shop has page types a service site does not. Each one wins a different kind of search. Matching searches to the right page type is most of the exercise. Getting it wrong is why so many product pages never appear for anything.
A Shop Has Page Types A Service Site Does Not
A service business has service pages and articles. A shop has categories, subcategories, product pages, filtered listings, brand pages and guides. Each type suits a different kind of search, so the research question is not only which searches to target but which sort of page should answer each one.
Why that changes the exercise. An extra decision.
On a service site you decide the subject and write a page. Here you decide the subject, then decide what kind of page it belongs on. That second choice decides whether it works.
The mismatch we see most. Product pages aimed too broadly.
A shop optimising individual product pages for searches describing a whole category. Those searches want a choice of options and a single item cannot provide one.
What follows from the structure. An order of priority.
Categories first, then the guides, then products. Most shops work in the opposite direction because products feel like the thing being sold.
The advantage a shop has. Existing structure.
Your catalogue already groups things. Half the clustering work is done. The job is checking whether your groupings match how people search rather than how you stock.
The disadvantage. Scale.
Thousands of pages, most generated automatically. That creates the specific problem in block seven, which is the one that quietly damages shops.
Category Pages Do The Heavy Lifting
Most people search for a type of thing rather than a specific item. That means category pages carry the commercial weight on almost every shop. They are usually the least considered pages in the whole catalogue.
Why people search that way. They are choosing.
Somebody who already knows exactly which item they want frequently goes straight to a retailer or types a model number. Everybody earlier in the decision is describing a type.
What that means for the page. It has to be a page.
Not a bare grid of products with a heading. Categories that win searches explain what the group covers, how to choose within it and what the differences mean.
What most category pages are. Empty listings.
A title, some filters and a grid. Nothing that helps somebody choose and nothing that distinguishes it from every competitor's version of the same page.
Where the content should go. Around the products.
An introduction that orients somebody, plus a fuller explanation below the listing for those who scroll. Neither should push the products down the screen.
The subcategory question. Follow the searches.
Create a subcategory where people genuinely search for that narrower thing. Creating them to tidy your catalogue produces pages nobody was looking for.
How to check yours. Read them as a stranger.
If a category page tells somebody nothing they did not know from the menu, it is a listing rather than a page.
Product Pages And What They Can Win
Product pages win searches naming that exact product. That is a genuinely valuable set of searches and a much smaller one than most shops assume. Expecting every product page to attract search traffic sets up a disappointment that is nobody's fault.
What they realistically win. Exact names.
The product name, the model number, the manufacturer code and close variations of those. People typing them are usually close to buying.
What they will not win. Category searches.
A single product cannot satisfy somebody comparing options, because comparison requires options. That search belongs to the category page above it.
Why most product pages get no traffic. Nobody knows the name.
If a product has no established name that people search, no amount of optimisation creates the demand. That is normal and it is not a failure.
What product pages are actually for. Converting.
Their job is turning somebody who arrived through a category into a purchase. Judge them on that rather than on search visibility.
The exception worth the effort. Products people name.
Well known items with recognisable names deserve individual attention, since those searches exist and are commercially strong.
The description problem. Manufacturer text.
Copying the supplier's description gives you a page identical to every other retailer's. Writing your own is what distinguishes it, which matters most on the products worth ranking.
Modifiers Are The Whole Game
The same product word with a different qualifier attached is a genuinely different search, made by a different person, wanting a different page. Modifiers do more work in retail research than anywhere else. They are where the winnable opportunities sit.
Attribute modifiers. The largest family.
Size, colour, material, capacity, finish. Somebody adding one of these has narrowed their requirement and knows what they want.
Use modifiers. The most revealing.
What the thing is for. A product described by its purpose tells you the situation somebody is in, which is far more useful than the product noun alone.
Audience modifiers. Frequently missed.
Who it is for. These create genuine subcategories and shops regularly have the stock without having the page.
Price modifiers. Two directions.
Cheap, budget and affordable at one end, professional and heavy duty at the other. Both are strong intent signals and they want completely different pages.
Comparison modifiers. Late stage.
Best, versus and alternative. Somebody is choosing between named options, which is block six.
What to do with them. Check the results.
For each promising modifier, search it and see whether category pages or product pages rank. That tells you which to build, per our intent guide.
The opportunity in them. Less contested.
Large retailers cover the broad product terms. Specific modifier combinations are frequently uncovered. A smaller shop can own those.
Brand And Model Searches
Somebody typing a manufacturer name and a model number has done their research and is looking for somewhere to buy. The intent is as strong as it gets. The difficulty is that everybody else selling the same item is competing for exactly the same words.
Who you are competing with. Everybody.
The manufacturer, the large retailers and every other stockist. All with the same product name and frequently the same supplied description.
What decides it. Something extra.
Stock availability, delivery, a genuine description, fair comparison against alternatives or expertise about the product. Differentiation rather than optimisation.
The brand page question. Sometimes worth it.
A page covering everything you stock from one manufacturer catches people searching the brand alone. Worth building where you carry a real range.
When it is not. Two products.
A brand page listing a handful of items is a thin page competing with your categories. The threshold is whether it is genuinely a useful destination.
The variation worth catching. Misspellings and part codes.
People type model numbers in several forms. Handling those variations is worth more than most optimisation, since each one is somebody ready to buy.
The trap. Brands you do not stock.
Building pages for manufacturers you cannot supply attracts people you cannot serve. It damages trust immediately when they discover it.
Buying Guides And Comparisons
People deciding what to buy search for help deciding. Those searches look informational and behave commercially, which makes them the most undervalued opportunity on most shops.
What they look like. Questions about choosing.
Which type suits what, what size somebody needs, what the difference is between two things, what to look for. All of them come from somebody preparing to spend money.
Why shops avoid them. They look like blogging.
A retailer focused on selling sees a guide as a distraction. In practice it catches people earlier and brings them into the shop already advised.
What makes them work. Genuine help.
Recommending against a purchase where it does not suit, being clear about limitations and answering the question fully. A guide that only pushes the most expensive option is transparent, plus useless.
The link that matters. Guide to category.
Somebody who has decided needs an easy route to the relevant listing. One clear link where a decided reader would look for it.
The comparison version. Two named things.
People compare specific options constantly. Handling those fairly, including saying when one is better, earns more trust than any claim about your own range.
Where these sit in the plan. Second.
After the categories and before the individual products, since they feed the categories directly.
The Filters Trap
Every filter combination a shop can generate is a potential page. Almost none of them should be one. Left unmanaged, a modest catalogue produces thousands of near-identical addresses, all competing with each other and with the category above them.
How it happens. Automatically.
Filters generate addresses. Combine three or four filters across a catalogue and the number of possible pages becomes enormous without anybody deciding to create them.
Why it damages a shop. Self competition.
Dozens of variations of the same listing divide whatever standing the category had. Our cannibalisation guide covers the mechanism.
The second cost. Attention wasted.
Search engines spend effort working through thousands of pointless variations instead of the pages that matter. On a large catalogue that is a real cost.
Which combinations deserve a page. The searched ones.
A filter combination people genuinely type is a legitimate subcategory. That is a small number and it is discoverable through research rather than assumption.
How to decide. The same test.
Does anybody search this exactly. Would this page serve them better than the category. If both answers are no, it should exist for shoppers and not as a separate page.
Who fixes it. Your developer.
The controls for this are technical rather than editorial. The research decides which combinations to promote and somebody else implements the handling.
What Not To Build A Page For
Restraint matters more on a shop than anywhere else, because the platform makes creating pages so easy. Every page that should not exist dilutes the ones that should. Shops accumulate these faster than any other kind of site.
Filter combinations nobody searches. The largest source.
Block seven. The default position should be that these are not separate pages unless research says otherwise.
Near duplicate categories. Common.
Two categories describing the same thing in different words, usually created at different times by different people. One should absorb the other.
Seasonal pages left standing. Quietly accumulating.
Last year's promotional listing, still live and still competing. Either update them annually or remove them properly.
Brands and lines you no longer carry. Trust damage.
Pages for stock you cannot supply. Point them somewhere useful rather than leaving somebody to find out at the basket.
Guides with no route to a purchase. Half a job.
Advice pages that never connect to the relevant listing. Useful reading, no commercial return.
How to know which pages earn their place. Measure it.
Ecommerce tracking shows which pages actually contribute to sales. Our ecommerce tracking guide covers setting that up properly.
Where the rest sits. The hub.
All twenty-two guides are on the keyword research guide.
Your category
pages are doing
the selling and
nobody wrote
them.
Most people search for a type of thing rather than a specific item, so the categories carry the commercial weight. On most shops they are a title, a row of filters and a grid, identical to every competitor's version of the same page. Meanwhile the effort goes into product pages aimed at searches a single item cannot win.
How we approach a shop:
Most product pages will never be found through search directly. That is normal rather than a failure. Judging them on it wastes years.
Every guide.
One subject.
What the words mean, what the numbers in a keyword tool actually are, how to find the searches worth targeting, how to turn a list into a publishing plan and what to do when a page you relied on slips.
