What an unhealthy structure looks like
The Ecommerce Crawl shows collection problems indirectly. You rarely get a single error called "bad
architecture"; instead you see patterns that point to it:
- Many collection pages with almost the same product list, differing only in order or a handful of items.
- Products that belong to a long list of collections, sometimes a dozen or more, while other products
belong to none and end up as orphans.
- Important categories that take four or five clicks to reach from the home page.
- Collections with one or two products, or none at all after stock sold out.
- In Search Console, several of your collection URLs appearing in turn for the same query, each with
modest impressions, rather than one page building a strong position.
That last pattern is often called cannibalisation. It is not a penalty; it is search engines being
unsure which of your pages best answers the query, so none of them gets the full benefit of your links
and relevance.
Give every collection one job
Most shop collections fall into one of four types. Mixing them up is the root of most overlap.
| Type | Examples | Purpose | Index it? |
|---|
| Core category | Walking Boots, Oak Chopping Boards, Waterproof Jackets | The main
landing page for a product type people search for | Yes, with its own introduction |
| Audience or attribute cut | Women's Walking Boots, Wide-Fit Boots | A subset with
genuine, distinct search demand | Yes, if demand and product count justify it |
| Merchandising collection | Sale, New In, Bestsellers, Under £50 | Helps shoppers
browse; content changes constantly | Case by case; they rarely rank for product terms |
| Campaign or seasonal | Christmas Gifts, Back to School | A recurring occasion with
its own searches | Yes, at a permanent URL |
A core category should be the one page that answers "I want to buy this kind of thing". Audience and
attribute cuts sit beneath it and only exist where people really search that way. Merchandising collections
are for browsing, and it is fine for them to overlap heavily with core categories, as long as they are
not written and linked as if they were competing for the same searches. Seasonal ranges are covered in
Fixing seasonal product handling.
💡 A useful test for any collection: finish the sentence "Someone searching for ___
should land here." If you cannot fill the gap with a real phrase, or the phrase already belongs to another
collection, the collection is not a search landing page.
Deciding between overlapping collections
When two collections target the same search, choose one to be the landing page and change the role of the other. Your options:
- Merge. If two collections are the same thing under different names, such as "Chopping
Boards" and "Cutting Boards", keep the one with more links and traffic, move all products into it, and
redirect the other.
- Differentiate. If both have a real purpose, such as "Walking Boots" and "Waterproof
Walking Boots", make sure the second is a true subset with its own introduction and products, and link
from the parent to it.
- Demote. If one is a merchandising collection, such as "Boots Under £80", keep it
for browsing but do not put the core search phrase in its title and heading, and link to the core category
from it.
- Consolidate signals with a canonical. For collections that must exist separately
but are near-identical in content, a canonical to the main collection can work. Use this sparingly; if
the pages are meant to differ, a canonical hides the one that differs.
For filter-generated versions of a collection, such as colour or size filters, the rules are different
and are covered in Fixing faceted URL
explosion.
Primary categories for products
A product can sit in many collections, and that is good for browsing. But each product should have
one primary category: the one that appears in its breadcrumb, that the product page links back to most
prominently, and whose URL path it uses if your platform puts categories in product URLs.
Consistent primaries matter because breadcrumbs and "back to category" links are internal links on
every product page. When a thousand boots all link back to "Walking Boots", that category receives a
strong, clear signal. When they link back to whichever collection the shopper arrived from, the signal
is scattered across Sale, New In and Summer.
Mark up the breadcrumb so search engines read the hierarchy the same way shoppers see it:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "BreadcrumbList",
"itemListElement": [
{"@type": "ListItem", "position": 1, "name": "Footwear", "item": "https://www.example.co.uk/footwear"},
{"@type": "ListItem", "position": 2, "name": "Walking Boots", "item": "https://www.example.co.uk/footwear/walking-boots"},
{"@type": "ListItem", "position": 3, "name": "Ridgeway Leather Walking Boot"}
]
}
</script>
Keep product URLs themselves independent of categories where your platform allows it. A product at
/products/ridgeway-leather-walking-boot can move between categories without its address
changing; one at /footwear/walking-boots/ridgeway-leather-walking-boot either changes URL
or keeps a misleading path. Duplicate addresses created by category paths are handled in Fixing
duplicate product URLs.
Depth and breadth
There is no magic number of levels, but the principle is simple: the pages you most want found and
ranked should be reachable in few clicks from the home page, through ordinary HTML links. Pages buried
five or six levels deep get crawled less and are seen as less important.
- Too deep: Home, Clothing, Outdoor, Men, Jackets, Waterproof, Lightweight. Many of
those levels have few products and little search demand of their own.
- Too flat: sixty top-level menu items with no grouping, which spreads link value thinly and confuses shoppers.
- Usually right: two or three levels for most catalogues, with extra levels only where
each has distinct demand and enough products to be useful.
Pagination is part of depth. If a category has forty pages of products, items on page thirty are a
long way from the home page. Better sub-categories, sensible default sorting and links from related products
all help, while very large catalogues also need the approaches in Fixing
crawl budget waste.
Empty and thin collections
Collections with one or two products, or none, rarely serve a searcher well. Decide what each one
is for. If it has search demand but few products today, keep it and write a useful introduction; if it
was created for a campaign long finished, merge it into its parent and redirect. An empty collection
returning a normal page with "no products found" can be reported as a soft 404, so do not leave such
pages live and linked in your menu.
⚠️ Do not delete collections in bulk as a clean-up. Some will have external links
or rankings. Check Search Console and your link data first, then merge and redirect rather than removing
outright.
The fix, step by step
Step 1
Map every collection and what it
targets
Export all collections or categories from your admin with their URL, product count and
parent. Add clicks, impressions and main queries for each from the Search Console Performance report,
and the number of external links if you have link data. Lay them out as a tree in a spreadsheet or mind-map.
The Ecommerce Crawl output shows every collection URL your shop exposes and surfaces orphan products
that no collection links to at all.
Step 2
Classify each collection by its
job
Label every collection as a core category, an audience or attribute cut, a merchandising collection
or a seasonal range. For each core category and cut, write the search phrase it should answer. Use a
keyword tool and Search Console queries to confirm people really search that way. Anything you cannot
attach to a genuine phrase is a browsing aid, not a landing page, and should be treated accordingly.
Step 3
Resolve collections that compete
for one search
Where two or more collections share a target phrase, choose the one with the best
history and content as the landing page. Merge true duplicates into it and redirect the others, rewrite
subsets so they target a narrower phrase with their own introduction, and remove the core phrase from
titles and headings of merchandising collections. Link each subset and merchandising page up to the main
landing page with clear anchor text.
Step 4
Assign a primary category to every
product
Add a primary category column to your product export and fill it for every product, choosing
a core category or subset, never a Sale or New In collection. Configure your platform or SEO extension
to use that category in breadcrumbs and in BreadcrumbList markup. Where the platform builds product URLs
from categories, switch to category-free product URLs if you can, planning redirects for the old paths.
Step 5
Flatten the path to important pages
Check
how many clicks each core category is from the home page by clicking through the menus yourself or checking
link depth in a site crawler. Remove levels that have little demand and few products, and link your most
important categories from the main menu or home page. Keep menus in plain HTML links that load without
scripts, so crawlers see the same structure shoppers do.
Step 6
Restructure with redirects, not
deletions
Build a mapping sheet of every collection URL that will change or disappear and its
new destination, choosing the closest matching collection each time. Add 301 redirects for all of them
in one release, update the menu, breadcrumbs, internal links and sitemap to the new addresses, and avoid
chains from earlier restructures. Recrawl straight after launch to catch any old collection URL still
returning 404 or linked internally.
Step 7
Set rules for creating new collections
Write
a short rule for whoever adds collections: which type it is, what phrase it targets, which existing collection
it must not compete with, and whether it is indexed. Keep merchandising and campaign collections out
of the main navigation unless they are permanent. Review the tree every quarter and after range launches,
merging anything that has drifted into overlap before it starts splitting your rankings.
Questions and answers
1. Is it a problem if a product appears in many collections?
Not by itself. Shoppers benefit from finding a product under Boots, Sale and Gifts
for Him. The problems start when those collections compete for the same searches, when the product's
breadcrumb and back-links change depending on the route, or when the platform creates a separate product
URL for each collection. Give each product one primary category for breadcrumbs, keep product URLs independent
of collections, and let the other collections exist purely for browsing.
2. Should Sale and New In collections be indexed?
They can be, but they rarely win searches for specific product types, and their contents
change constantly. Many shops keep them indexable because people do search for a brand name plus sale,
but avoid giving them the same titles and headings as core categories. If a sale collection is only live
for a few weeks, consider keeping one permanent sale URL rather than creating a new one each time, so
any value it gathers is kept.
3. How many levels deep should my category structure be?
There is no fixed rule. Most shops work well with two or three levels, adding a level
only where it has distinct search demand and enough products to be useful. What matters is that your
important categories are a few clicks from the home page through ordinary links, and that each level
helps the shopper narrow down. Levels that exist only for tidiness, with few products and no searches,
usually add depth without adding value.
4. What is keyword cannibalisation between categories?
It describes several of your own pages targeting the same search, so search engines
swap between them or split relevance and links across them. On shops it often happens when a core category,
a similar sub-category and a merchandising collection all use the same phrase in their titles. It is
not a penalty, but it usually means none of the pages ranks as well as one clear landing page would.
Merging or differentiating the pages fixes it.
5. Should my product URLs include the category?
It is usually safer if they do not. A category-free product URL stays the same when
you move the product between categories or restructure your catalogue, and it avoids one product being
reachable at several category paths. Category paths in product URLs can look tidy, but they tie the address
to a decision that often changes. If your shop already uses them and ranks well, changing them is a migration
that needs full redirects, so weigh it carefully.
6. Can I use canonical tags to stop categories competing?
Sometimes, for collections that really are near-identical and must both exist, for
example one required by a marketplace integration. But a canonical tells search engines the pages are
duplicates, so it is wrong for collections that are meant to be different, such as a parent category
and a narrower sub-category. For those, differentiate the content and link between them. For collections
that are simply the same thing twice, merging with a redirect is cleaner than a canonical.
7. Will restructuring my categories hurt rankings?
Any large structural change carries some short-term risk, because search engines need
to recrawl and reassess the new URLs. The risk is much smaller when every old collection URL redirects
to its closest new equivalent, internal links and sitemaps are updated at the same time, and you avoid
changing product URLs unless you must. Make the change outside your busiest season, recrawl straight
away to catch broken links, and watch Search Console closely for several weeks.
See how your collections overlap
The Ecommerce Crawl maps the URLs on your store, finds duplicate product URLs and faceted navigation
getting out of hand, surfaces orphan products and measures how efficiently your crawl is spent.
Open Ecommerce Crawl Fixes →