Ecommerce Site Structure for Categories and Filters

What ecommerce site structure decides

Set the number of category levels, the links at each level, each page's URL, and the filter combinations that need their own pages. All four are easier to set now than to change once the catalog has grown.

Later changes can involve new URLs and a wait for Google to recrawl the affected pages.

Google works out how your pages relate from the links between them. Its ecommerce documentation identifies two signals: the number of links Google follows to reach a page, and the number of internal links pointing at it.

The more internal links a page has, the higher its relative importance inside your site.

Putting more links at each level shortens the route: the right-hand tree reaches the same products in three clicks instead of five.

For example, a lighting store could organise 1,296 products into 36 ranges without a third level.

Baymard Institute scored 13,000 homepage and category navigation elements across 344 top-grossing US and European stores.

It rated 76 percent of those stores mediocre or worse.

How Google reads your site structure

Googlebot finds your products by following links through the category tree. Google's ecommerce documentation recommends links from menus to category pages, from category pages to subcategory pages, and from subcategory pages to every product page.

If your category pages do not link directly to every product, Googlebot might not find all of them by crawling alone. Google strongly recommends linking to every product you want indexed.

If you cannot link to every page, submit the remaining URLs through a sitemap or a Google Merchant Center feed.

A feed entry does not increase a page's relative importance; that comes from internal links.

For example, say you sell 1,400 taps. If your range pages show 24 products and never link to page two, the other 1,376 taps exist only in your sitemap.

Use an internal linking map to find the pages that need more links or a shorter path from the home page.

How many levels should an ecommerce site have?

Google does not prescribe how many category levels sit between the home page and a product. When you choose the number of levels, consider how quickly shoppers can reach a product and how many links Googlebot must follow.

Panayiotis Zaphiris asked 17 people to find pages in five link hierarchies, each built over the same 64 destination pages. The result appeared in the Human Factors and Ergonomics Society proceedings in 2000.

In Zaphiris's experiment, participants took about 20 seconds to find a page in the two-level hierarchies.

Participants took 36.0 seconds in the six-level hierarchy and rated it the hardest to use. The time needed to find a page differed significantly across the five hierarchies, but the number of errors did not.

Zaphiris found that adding a level slowed people down, while increasing the links at a level from two to sixteen did not change how quickly they found a page.

Google counts the links it follows to reach a page, so adding a category level lengthens the route from the home page to every product beneath it.

For example, a store with separate levels for range, material and finish can usually combine material and finish into one. Shoppers then reach the same products with one fewer click from the home page.

How many links should each level of the site have?

Most category trees use fewer links per level than they could. In Cyrus Shepard's study the association between internal link count and Google clicks was strongest at around forty links per page, and it weakened above about fifty.

Cyrus Shepard compared internal link counts with Search Console clicks across 1,800 websites and about 520,000 URLs. The study covers 23 million internal links and was last updated on 23 February 2026.

URLs with 0 to 4 internal links averaged two Google clicks. URLs with 40 to 44 internal links averaged four times as many.

Above roughly 45 to 50 internal links, average Google clicks decline again. Shepard attributes much of that decline to links in sitewide navigation, which repeat on every page and often point at URLs nobody visits.

The HTTP Archive measured a median of 43 internal links per desktop page across 16,213,084 websites in its July 2025 crawl.

A 400-item mega menu far exceeds both numbers: the median of 43 internal links per desktop page, and the 40 to 44 band with the highest average Google clicks.

Google states that Googlebot crawls the first 2MB of a supported file type, and your menu's code sits inside that 2MB. For example, on a store with 40,000 URLs, a footer containing 600 links is fetched 40,000 times.

Give each page on the site one address

When a page has two URLs, a search engine may select a different one from the URL your internal links use. A product listed in several collections has more than one URL by default.

Google recommends readable, descriptive words in the URL path instead of long ID numbers, and hyphens instead of underscores. It also asks you to trim parameters that do not change the content.

Should the category appear in the product URL?

Shopify's Liquid documentation describes two URLs for the same product content: a standard product URL and one that includes a collection path. Shopify advises theme developers to consider the SEO implications before using the filter that generates them.

Adobe Commerce leaves the category path out of product URLs by default. When a store turns the path on, the canonical URL still points at the product URL key.

So open a category page, inspect a product link, and compare it with the canonical on the product page. A canonical tells Google which address you prefer, and an internal link is the address Googlebot follows.

Before you plan a URL change, check which parts of the URL pattern your platform lets you change. Shopify fixes parts of it, including /products and /collections.

For example, if your collection template uses product URLs that include the collection path, every category page built from it links to a non-canonical URL.

Decide which filters become pages on the site

Filter combinations can generate more URLs for one category page than the catalog has products. Google's faceted navigation documentation says filters can create a near-endless number of addresses, and that crawl time spent on them is time not spent on new pages.

Check whether anyone searches for each filter combination. According to Ahrefs, 99.84 percent of keywords receive fewer than 1,000 monthly searches, and together they account for 39.33 percent of total search demand.

A low monthly volume is not a reason to skip a filter page.

Skip it when nobody searches the combination, or when the page would repeat what the parent category already says.

For filter pages you want crawled, follow Google's URL guidance. Use the standard ampersand as the parameter separator, keep the filter order consistent, prevent duplicate filters, and return a 404 for combinations with no results.

Search Engine Journal reported in December 2024 that Gary Illyes identified faceted navigation as the leading cause of overcrawling problems reported to Google.

Count the URL combinations each filter group can produce, then decide which of them Google should crawl.

For example, a 300-product category with color, size and price filters can produce more addresses than it has products. If none of those combinations has search demand, do not build pages for them.

On Shopify each collection has its own URL, so this decision happens when you create collections. Check the collection settings when you decide which filter combinations get their own URLs.

What changing the structure later costs

Setting the platform options and the number of category levels costs less while the catalog is small. Later changes require redirects and time for Google to recrawl, and Shopify prevents changes to its fixed URL paths altogether.

Google says a medium-sized site move with URL changes can take a few weeks or more before new URLs start appearing.

It asks you to keep redirects for at least a year.

Shopify states that you cannot redirect URLs that use its fixed paths: /products, /collections and /collections/all.

For example, Shopify does not allow you to move a catalog from /collections/taps to /taps.

In Adobe Commerce, you can enable automatic redirects when a URL key changes.

Treat a structure change that alters URLs as a site migration. Map the old URLs to the new ones, redirect them, then wait for Google to recrawl.

If you would rather not run that sequence yourself, our SEO migration service plans and carries out the move.

Measure the structure you already have

Use Search Console or a site crawler to record four things about your current structure. Record a baseline before you make changes, then repeat the measurements a few weeks later and compare the results.

Google says crawl budget deserves attention on sites above a million unique pages that change weekly, or above 10,000 pages that change daily. It calls both figures a rough estimate.

So a 900-product store is almost never short of crawl budget.

Focus instead on whether Google can find your pages, and on how internal links signal their relative importance.

Count the products at each click depth instead of relying on the average depth. A five-click path affecting 8,000 products deserves more attention than a seven-click path affecting one.

For example, say 70 percent of your products sit four hops from the home page and your bestsellers sit at two. The shorter paths to your bestsellers signal their importance to Google.

Common ecommerce site structure mistakes

Check for four mistakes that add clicks to your product paths or create duplicate URLs. Each one shows up in the four measurements above, and each one is cheaper to fix before a rebuild than after.

The first is adding a category level to solve a labelling problem. If shoppers cannot tell two categories apart, rename them and rewrite the page copy, because a new parent level moves every product further away.

The second is creating a page for a filter combination because the platform allows it. Check search demand first, using one keyword lookup per filter group.

The third is treating a sitemap as a substitute for internal links. A sitemap helps Google find a URL; relative importance still comes from internal links.

The fourth is letting the theme link to product URLs that differ from the canonical you chose. Compare a category page's product links with the canonical on the product page.

For example, adding a Brands category above the existing ranges would increase the path to 6,000 products from three clicks to four.

Record these metrics before you rebuild, so you can compare them afterwards against the rest of your ecommerce SEO work.

Start with one branch of the site

Test the changes on one category branch before you rebuild anything else. Choose the category branch with the most revenue. Follow the five steps below and compare the numbers a month later.

1. Count the clicks to a best-selling product

   Click from your home page to one of your best-selling products and count the clicks. Do it for five products in the same branch.

2. Find two levels you can combine

   If two levels in that branch group products by attribute instead of by the categories shoppers search for, write out the combined category names. Check each name for search demand.

3. Check which address your template links to

   Open a category page, inspect a product link, and compare it with the canonical on the product page. On Shopify, compare the URL that includes the collection path with the plain product URL.

4. List the filter combinations with demand

   Check search demand for each filter combination in that branch. Keep pages for the combinations with demand and plan to block crawling of the rest.

5. Record the four measures and the date

   Record the click depth, the number of internal links per URL, the canonical Google selected and the indexed filter count. Repeat all four checks a month after the change and compare the results.

If products take fewer clicks to reach and internal links point to your chosen URLs, check the other branches for the same problems before you combine their levels.

We run this check first in the stores we work on as a search marketing agency.