A category tree works for search when every product you want indexed can be reached by following ordinary links from the menu, through a category, to the product. Google reads the structure of an online store from those links, not from the shape of its URLs. If you run or build a marketplace with hundreds of sellers, three decisions matter most: how the tree branches, how deep it goes, and what happens to products that appear on no category page at all.
Picture a marketplace for used and new film camera lenses. Sellers list prime lenses, zooms, adapters and lens caps. Buyers search by mount, by focal length and by brand. That mix is exactly the kind of catalog where menus sprawl, subcategories stay half empty and some listings can be found only through the on-site search box.
What Google's guidance says about links and store structure
According to Google's page on ecommerce site structure (Opens in a new tab), Google tries to find the best content on a site by looking at how pages relate to one another through links. Navigation elements such as menus and cross-page links influence how Google understands the structure. The page adds that Google can use link-based signals, including how many links it has to follow to reach a page and how many links point to that page, to estimate how important the page is relative to the rest of the site.
Two practical readings follow. A listing that sits several clicks from the home page and receives few links may look minor next to the rest of the store. And the page phrases this as something Google "can" use, so it is a risk to manage, not a published ranking formula.
The chain from menu to product
The recommended build is a hierarchy of links: menus link to category pages, category pages link to subcategory pages, and subcategory pages link to all product pages. The page also asks that every page be reachable by following links through the site's navigation. The phrase "all product pages" carries the weight. The page states that category pages that do not link directly to every product in the category can keep Googlebot from finding all products.
That sentence is why a lens subcategory that shows the first group of listings and hides the rest behind an interaction with no real link is a structural gap, not a cosmetic one. Pagination and incremental loading belong to a separate decision, covered in the pagination guide. The principle here is narrower: each product needs at least one crawlable link from somewhere in the tree.
Why URL patterns do not draw the tree
The same page says Google does not generally infer a site's structure from URL patterns; it analyzes links between pages to judge relative importance. Two misreadings follow from ignoring that. A URL that contains a category word does not prove a tree exists. And counting slashes in an address says little about depth. A lens listing with a flat address can sit three clicks from the home page, while a listing with a long path may be reachable in two.
URLs still matter, for a different reason. Google's guidance on ecommerce URL structure (Opens in a new tab) says well-designed URLs help Google crawl and retrieve ecommerce pages. Descriptive words in the path, one consistent letter case, key=value query parameters and no repeated parameters all serve crawling. Whether product addresses should carry the category path is a separate question, and the documentation has no statement on it.
Designing the tree for a film lens marketplace
Before drawing branches, answer three questions. What word does a buyer type for this listing? If one lens fits several categories, which one is its home? Will the category be split when it grows, or stay a single page? Subcategories opened without answers to these are merged later, and every merge leaves redirects to maintain.
Example scenario: lens categories
Example scenario: the top menu has one entry, "Lenses". Under it sit "Prime lenses", "Zoom lenses" and "Adapters and accessories". "Prime lenses" splits by mount into "M42", "Canon FD" and "Nikon F". A 50 mm lens with an M42 mount belongs under "Prime lenses" and "M42" at once. A lens-and-adapter bundle could also sit under "Adapters and accessories". And the "Nikon F" subcategory currently holds only a handful of listings.
| Level | Example page | Links to | Question for the manager |
|---|---|---|---|
| Top menu | Lenses | All subcategories | Is this the broad word buyers use? |
| Subcategory | Prime lenses | Mount groups and listings | Do the listings justify a split? |
| Subcategory, second level | M42 | Every listing in that mount group | Is each listing linked with a crawlable anchor? |
| Product | A 50 mm M42 lens | Parent category and related listings | Which category is its home? |
A home category for every listing
Practical suggestion: store one "primary category" per listing. The listing can still appear elsewhere, but breadcrumbs, canonical choices and sitemap grouping follow the primary one. With that field in place, a lens that fits both a mount group and a bundle category still has one agreed way of being described. Breadcrumb mechanics are covered in the guide to site architecture and breadcrumbs.
Depth is a measurement problem, not a number
How many levels are enough? Google's pages give no number. The page gives a pair of principles: the number of links needed to reach a page can feed an estimate of importance, and every page should be reachable through navigation links. A depth decision therefore needs your own measurements.
Three measurements to track
Practical suggestion: track click distance (the number of links followed from the home page to a listing), the number of internal links each listing receives, and the share of a category's listings that appear as direct links on the category page. A crawling tool or a short script can produce all three. The page does not ask for these figures, but each one turns an abstract sentence into a question you can answer.
Decision rule: open a subcategory or keep one page
Practical suggestion, as a rule of thumb: before opening a subcategory, ask these four questions. Open it only when at least three answers are yes.
- Do buyers search for this group with a distinct term?
- Are there enough listings that browsing the group is useful?
- Can the group carry its own short description?
- Can sellers tag listings into the group consistently?
The fourth question matters in a marketplace, because sellers enter the data. If tagging is inconsistent, the subcategory ends up half empty. What to do with empty or thin category pages is a separate decision with its own options: noindex, merging or a 404 status.
Back to the scenario. If "Nikon F" has only a handful of listings and few buyers search for it as a separate term, the manager can keep it as a filter on the "Prime lenses" page. If buyers do search for it, it can stay a page. Nothing in the documentation settles that choice for you.
Products that only search can reach
Google's page names this problem directly: products reachable only through a search box may be missed, because Googlebot generally does not submit searches into search boxes while crawling. The page itself hedges with the word "generally", so treat it as a risk, not a law. The design principle is plain anyway: link to every product you want indexed.
In a marketplace the gap opens at upload. A seller creates a listing, it enters the search index, and a buyer who types the right words finds it. If the category page does not list it, or no category was assigned, the listing is reachable only through search. Call it an orphan listing. Orphans are found by an internal audit, not by looking at the site.
Finding orphan listings
Practical suggestion: compare two lists. The first holds every published listing in the database. The second holds the product URLs a crawler finds when it starts at the home page and follows links only. Anything on the first list and missing from the second is an orphan.
| Cause | Symptom | Direction of the fix |
|---|---|---|
| No category assigned | Listing exists in search and at its direct address only | Make category selection mandatory on the upload form |
| Category query leaves it out | Category count and listed items differ | Review the filters in the category listing query |
| Deep pagination only | Only a far page links to the listing | Review default sort order and page links |
| Link works through a script event | No href in the page source | Use a real anchor element |
Crawlable links in menus, cards and pagination
The page says to use a elements with an href attribute and not to rely on JavaScript events attached to other DOM elements for navigation. The URL structure page repeats this and notes that Googlebot may not detect navigation built with JavaScript. It also asks for meaningful anchor text, such as a product title, and names "click here" as the kind of generic text to avoid.
A lens card built as a clickable container whose click handler loads the listing shows the product to a visitor but offers no link to a crawler. The fix is to wrap the title, or the card, in a real anchor. Check menus, category cards and pagination controls for the same pattern.
Checklist for menus and cards
- Is every main menu entry an anchor with an href?
- Does the dropdown carry its category links in the source without scripts running?
- Does each listing card use the product title as link text?
- Do pagination links contain real hrefs?
- Does each subcategory page receive a link from its parent?
- Do breadcrumb items link to category pages with real anchors?
Promoting key categories and listings
Google's page says it does not generally infer structure from URL patterns, and it adds a general rule: more internal links to a page signal greater importance relative to other pages on the site. It suggests linking to best-selling products from the home page and from other content such as blog posts or newsletters. Its closing note says the ultimate ecommerce SEO practice is to create useful, interesting content that gives value to users.
In a marketplace this is a decision for the platform team, not for individual sellers. Which categories sit on the home page, and which collections get editorial links, are strategic choices. A seasonal feature on "Prime lenses" could earn links from the home page and from articles about choosing a first manual lens. Those links should help a buyer, because the page's own closing advice is about value, not link counts.
Keeping promotion honest
Practical suggestion: keep showcase links on the home page and in content few and reviewed on a schedule, using sales data. A menu stuffed with hundreds of product links serves no reader, and the documentation does not suggest it.
Sitemaps and feeds as support, not substitutes
The page states that if linking to every page is not possible, a sitemap or a Google Merchant Center feed can point Google to pages a crawler would not otherwise find. The sentence frames both as backup for gaps in the link structure. Practical suggestion: do not let a sitemap hide a structural hole. A listing that appears in the sitemap and nowhere in the tree belongs on the orphan list.
How to split sitemaps for a large marketplace and how to use lastmod are separate technical topics. A Merchant Center feed is another way to give Google product data; managing category mapping inside a feed is its own task and does not replace the category tree on the site.
Sellers, forms and category quality
The manager draws the tree, but sellers fill it. Each seller chooses a category when uploading, and that choice decides which page links to the listing. A wrong choice puts a lens in the wrong context for buyers and for Google alike.
Designing the category field
Practical suggestion: use a managed list in place of free text, ask for the deepest relevant category, and show a one line description next to each. In the lens scenario, a seller unsure whether an M42 lens-and-adapter bundle belongs under "Prime lenses" or "Adapters and accessories" can read the descriptions and decide, using the listing's real nature as the test.
| Upload step | Effect on the tree | Suggested control |
|---|---|---|
| Category choice | Decides which page links to the listing | Required field, managed list |
| Listing title | Becomes link text on category pages | Check for descriptive, unique titles |
| Publishing | Listing appears on the category page | Automatic check that it is listed |
| Editing | A category change removes it from the old page | Change log for category edits |
Filters versus real categories
Buyers filter a category by focal length, aperture or price. Filter pages are not part of the category tree, but when they create URLs they affect crawling. Google publishes a separate page on managing the crawling of faceted navigation URLs, and the ecommerce guidance points to it. Handling those URLs is outside this topic.
Practical suggestion: do not confuse a filter result with a category. Ask whether a filter combination matches a distinct search intent. "M42 lenses" is a page buyers look for. A price range or a sort order is not a category. Drawing that line shows which pages in the tree are permanent and which are results.
| Page type | Place in the tree | Decision |
|---|---|---|
| Top category (Lenses) | Linked from the menu | Permanent page |
| Subcategory (M42) | Linked from its parent | Permanent if buyers search for it |
| Filter result (price range) | Inside a category, outside the tree | Handle with the faceted navigation guidance |
| On-site search result | Outside the tree | Cannot be relied on as a link source |
The listing grid on a category page
Google's page puts weight on category pages linking straight to the products in the category. In a large category, one page cannot list everything, so listings are reached through numbered pages, a "load more" button or infinite scroll. The URL structure page calls pagination a frequent source of URL errors and asks that each page in a paginated series have its own unique URL. Pagination choices are a topic of their own, covered in the guide to category pagination.
The tree question is concrete: how many steps separate the first page of a category from a listing on its last page? If pagination offers only a "next" control, the listing on the tenth page is nine steps further from the home page than the first page is. Practical suggestion: show numbered page links as real anchors, and order listings so that new and high value items sit on early pages. The ordering criterion is a business call.
Auditing infinite scroll and "load more"
These interfaces load listings with JavaScript. Google's page points to its JavaScript SEO basics and asks for real links for navigation. Practical suggestion: even if the interface scrolls endlessly, offer the same listing set in a paginated version with real links, then confirm that both views cover the same listings. To test, look at the page source with scripts off: do listing links appear?
A quarterly upkeep calendar: example scenario
Example scenario: the lens marketplace holds an upkeep meeting at the start of each quarter, and the output is a release note with four items.
- List categories with zero or very few listings and decide for each: keep, merge or remove.
- Run the orphan comparison and tell sellers which listings have no category.
- Refresh the categories and listings featured on the home page using sales data.
- Inspect menu and card links in the source for hrefs.
The calendar is invented for illustration. Its point is a rhythm that catches decay before it builds up.
Who owns what
| Owner | Area | Output |
|---|---|---|
| Platform manager | Tree structure, menu, featured items | Approved category list |
| Developer | Anchor elements, pagination, category template | Hrefs verified in the source |
| Content lead | Category descriptions, editorial links | Content calendar |
| Seller support | Category choice and upload warnings | Seller guidance |
Reading one listing end to end
Example scenario: a 50 mm M42 lens listing. From the home page you open "Lenses", then "Prime lenses", then "M42", then the listing: four steps. If the home page also shows a featured "Prime lenses" tile that links straight to the M42 group, a shorter path exists. Google's page suggests that the number of links a crawler has to follow to reach a page, and the number of links pointing to the page, can both enter the importance estimate. How they are weighted is not stated.
Practical suggestion: pick one listing per category and walk its path by hand. Click through the menu, find the listing on the category page, then check that the breadcrumb returns you to the category. Hand testing shows experience problems that a crawler hides, such as a listing buried deep in pagination or a mobile menu that holds a different set of links from the desktop one. When the two menus differ, the tree behaves like two separate structures.
When the tree breaks
Trees decay. Sellers remove listings, categories are renamed, and a manager reshuffles the menu for a campaign. Table: ways the tree breaks and how to watch for each.
| Break | Effect on the tree | How to watch |
|---|---|---|
| Category renamed, address changed | Internal links to the old address fail | List 3xx and 4xx responses in an internal link crawl |
| Subcategory emptied | A menu entry leads to a page without listings | List categories with zero listings every week |
| Menu trimmed for a campaign | Some subcategories lose their menu link | Measure click distance again after each menu change |
| New product type, no category | New listings become orphans | Run the orphan comparison with every release |
| Category template rebuilt | Listing links may move into scripts | Compare href counts in source with the previous release |
Practical suggestion: record every menu and category change in a release note and tie the link audit to that note. Whoever changes the tree should be able to measure what the change did to the link structure.
A thirty day audit plan
A month is enough to audit and repair an existing store. The plan turns Google's principles into jobs.
Week one: inventory
Export the full category list with listing counts. Check the primary category of each listing. Collect the listings with no category. Inspect the menu in the page source and note entries without an href. In the lens scenario you should end the week holding a table of every subcategory under "Lenses" and its listing count.
Week two: reachability
Crawl from the home page following links only and list the listings found. Compare with published listings. The difference is your orphan list. For each listing, record click distance and the number of internal links received.
Week three: repair
Assign orphans to categories or unpublish them. Decide on thin categories with the decision rule. Convert script-driven links into real anchors. Confirm that category pages link to every listing in the category.
Week four: re-measure and set a routine
Run the crawl again and compare with week two. Write the release note. Add a mandatory category to the upload form. Schedule a monthly orphan comparison.
Final checklist
- Every published listing is reachable from at least one category page through a crawlable link.
- Menu, category and card links carry hrefs.
- Each listing has one primary category.
- Thin and empty categories have a recorded decision.
- Key categories and best sellers receive links from the home page or from content.
- The sitemap supports the link structure and does not replace it.
Category page copy and heading structure sit outside this audit. For those, see the guide to ecommerce category page copy and structure.
Decision mistakes to avoid
Tying depth to a number
A fixed rule such as "stay within three levels" does not appear in Google's pages. Measure click distance, internal link counts and orphan listings.
Swapping the sitemap for links
The page offers sitemaps for cases where linking to every page is not possible. It does not remove the need to link category pages to products.
Improving search and leaving the link gap
A better on-site search helps buyers. It does not fix the crawlable link structure. Search and the tree are two systems that complement each other.
Frequently Asked Questions
How many levels should a category tree have?
The documentation has no statement on it. Google's page says the number of links needed to reach a page can be used to estimate importance. Measure click distance and internal links in place of fixing a level count.
Does Google read my store structure from URL folders?
According to Google's page, it does not generally infer site structure from URL patterns. It analyzes links between pages to judge relative importance.
Will products reachable only through on-site search be found?
Google's page says they may be missed, because Googlebot generally does not submit searches into search boxes while crawling. Link to every product you want indexed.
What if I cannot link to every page?
The page says a sitemap or a Google Merchant Center feed can point Google to pages a crawler would not otherwise find. They support the link structure and do not replace it.
Is a JavaScript click handler enough for a product card?
The page asks for a elements with an href and says not to use JavaScript events on other DOM elements for navigation. A card that works only through a script event cannot be relied on as a crawlable link.
How do I raise the profile of best-selling listings?
Google's page suggests linking to them from the home page or from other content such as blog posts or newsletters, and says more internal links signal greater importance relative to other pages on the site.
Conclusion
Three questions settle the category tree: can every listing be reached through crawlable links from the menu, is depth measured and not guessed, and does every product that buyers find through search also have a link path? Google's pages give principles and no level count: links carry an href, category pages link to products, and sitemaps or feeds back up the structure without replacing it. In the lens marketplace and in any other catalog, the order of work is the same: inventory, orphan comparison, repair, then a routine.
Once the tree is sound, you can look at how listings reach their audience. The Akua Adforce ad spaces page shows the placements available on the marketplace.
Post a comment
Please log in to leave a comment. Login