Mining Competitor Category and Filter Names

To mine a competitor's category and filter names, read its public menus, breadcrumbs and facet groups by hand, record each label with the URL words behind it, line up different names for the same concept, and test each candidate against your own catalog before it becomes a category. The result is a working vocabulary, not a copy of anyone's navigation. It suits marketplace and store teams that are about to name or rename categories and filters.

The running example is a used books marketplace, and its marketplaces, labels and tables are invented. Google's rules come from its Search Central ecommerce pages.

Why a category label is a keyword decision

A category name appears in three places at once: as link text in the menu, as the heading of the category page and as a word in the URL path. Choosing it therefore shapes how shoppers find the group and how a search engine reads the page. Google's ecommerce site structure page (Opens in a new tab) says Google tries to find the best content by analyzing the relationships between pages based on their linkages, and that navigation structures such as menus and cross-page links can affect how Google understands the structure of a site. For the text that supports a category name on the page itself, see ecommerce category page SEO copy structure.

The same page adds that Google can use information such as the number of links it needs to follow to reach a page, and the number of links to a page, to infer the relative importance of that page. A label in a menu is link text for the page behind it. Treat it that way when you read a rival's menu.

Labels as link text

Google's link best practices ask for anchor text that is descriptive and relevant to the page it points to, and give generic phrases such as "click here" as examples of weak text. The ecommerce URL page makes the same point for product links: include meaningful text between the anchor tags, such as the title of the product, and avoid generic phrases. Apply that to category labels. "Hardcover Novels" says what the page holds, while "Browse" or "More" does not.

Labels and URL words

Google's ecommerce URL structure page (Opens in a new tab) recommends descriptive words in URL paths, giving /product/black-t-shirt-with-a-white-collar as recommended and /product/3243 as not recommended. It also advises converting URL text to one case if the server treats upper and lower case alike. When you read a rival category, put the visible label and the URL word in two separate columns, since they do not have to match.

Collecting a rival's public navigation

Reading navigation is a manual review of public pages. It does not involve logging in, bulk downloading or any automation that breaks a site's terms of use. Work through the menu level by level.

1. Write down the top-level labels in the main menu.
2. Open each top-level label and record the sub-labels one level down.
3. Open sub-category pages and note the page heading and the breadcrumb labels.
4. Copy the URL path from the address bar for each label.
5. Record the facet groups and their option labels on a category page in a separate sheet.
6. Add the rival's name and the reading date to each row.

Keep the sheet to names

The sheet needs labels and URL words, not rival category intros, listings or photographs. Short names keep the table clean and remove any question about copied text.

Facet labels as attribute vocabulary

Facets on a category page show which attributes a rival uses to divide its stock. In a used books marketplace the groups are about the book and its copy: format, condition, language, edition and price band. Each group name and each option name is a candidate term.

The split helps because genre-type groups can form the skeleton of the category tree, while condition and language groups become filters. Whether filtered pages should be indexable is a separate decision, covered in the ecommerce filter URLs and faceted navigation guide. Google also keeps a separate document on managing the crawling of faceted navigation URLs and points to it from its site structure page.

Parameter names behind the facets

When selecting an option changes the address, add the parameter names to your sheet. Google recommends ?key=value parameters over a bare ?value, with /t-shirt?color=green as the recommended form and /t-shirt?green as the one to avoid. It advises against using the same parameter twice, since Googlebot may ignore one of the values, and shows ?type=candy,sweet as the recommended way to pass several values. It also asks you not to link internally to temporary parameters such as session IDs, tracking codes and values relative to the user, because they create short-lived or duplicate URLs. Record the first kind and skip the second: a session parameter is technical noise, not vocabulary.

Note whether each facet changes the address at all. A facet that filters the list in place is a term for shoppers only and has no URL word to record.

One concept, several names

Reading two or three rivals shows the same idea under different labels. One marketplace calls the group "Used Books", another "Second-Hand Books", a third "Pre-Owned Books". A synonym map puts these on one row. Its job is to show which labels point at the same group of listings, not to declare a winner.

The table is invented. In a real vocabulary the decision column comes from your own site search logs, buyer questions and Search Console queries. A rival menu cannot stand in for that data.

A rule for choosing between names

1. Do both labels lead to the same group of listings? If so, put them on one row.
2. Is one label a subgroup of the other? Then record it as parent and child.
3. Does your own stock include listings for both? If not, do not open a category for the missing one.
4. Which label do your own search logs favor? Take any numbers from your own data only.
5. Keep the unchosen label as a recorded synonym and use it naturally in category descriptions.

Fitting the vocabulary to your category tree

A vocabulary has to land on the nodes of a tree. Google describes the pattern of linking from menus to category pages, from category pages to sub-category pages and from sub-category pages to all product pages. If category pages do not link directly to all products, Googlebot might not find all products by crawling alone. Those products may be reachable from a search box, but Google writes that Googlebot generally doesn't try to submit searches into a search box as part of crawling a site. The page strongly recommends linking to all products you want indexed and, where that is not possible, using a sitemap or a Google Merchant Center feed.

A tree for the example

Example scenario: the vocabulary for the used books marketplace lands in a tree like this. The tree is invented.

• Fiction
  • Mystery and thriller
  • Science fiction
  • Historical novels
• Non-fiction
  • History
  • Cooking
  • Travel
• Collectible
  • First editions
  • Signed copies

Each node has one vocabulary row, one menu link and one URL in the sitemap. Condition and language stay as filters under each node and are not categories of their own.

Empty categories

Google advises avoiding links to, or at least indexing, pages without useful content. If a category has no items, it suggests the noindex robots meta tag. If your site detects that a category has become empty and automatically removes it from on-site search and browse, consider returning a 404 status code. A vocabulary row for a node with no listings is a reason to hold the category back.

One URL everywhere

Google recommends using the same URL in internal links, sitemap files and canonical tags. When the vocabulary changes a URL word, update all three together, so the menu does not point to one address while the sitemap lists another.

Spelling, plural and regional variants

English labels vary in several ways that a vocabulary should record: singular and plural (book, books), hyphenation (second-hand, secondhand), and regional spelling (hardcover, hardback). Give each variant its own column so that the menu, page headings and category descriptions stay consistent. How Google matches such variants is not stated in the ecommerce structure or URL pages, so treat the columns as a consistency tool, not as a ranking promise.

• Pick one visible label and one URL word for each concept and apply the same rule across the tree.
• Fix URL letter case to a single form, as Google advises when the server treats cases alike.
• Decide which variants appear only in descriptions and not in menus.
• Where a site serves several languages, keep one vocabulary per language. A rival's label in another language is no translation guide for your own.

Breadcrumbs and page headings as a second reading

Menus show how a rival wants shoppers to browse. Breadcrumbs and page headings show how the rival names a page once you are on it, and the two can differ. A menu may say "Cooking" while the breadcrumb on the category page says "Cookbooks and Food Writing" and the page heading says "Cooking Books". Record all three when they differ, in separate cells, because each is a distinct place where the term is used.

The disagreement tells you something about the rival's process. A menu label shortened for space, with a longer heading on the page, is a pattern you can use in your own template: a short label for the navigation bar and a fuller heading for the page. A mismatch that looks accidental, such as a menu term the page does not use, is worth a note but not worth imitating.

Reading the page title and the visible heading

Open the category page and compare three strings: the label you clicked, the heading at the top of the page and the browser tab title. If you need the tab title, it is visible without any tool. Write the three strings into one row. When all three use the same words, the rival keeps its naming consistent. When they differ, mark the row for review in your own vocabulary so that your team decides on purpose how much variation to allow.

This reading does not claim anything about how the rival performs. It only documents naming. The documentation has no statement on how a rival's title and heading choices affect its results.

How deep is the path to a listing?

Google counts the number of links needed to reach a page among the signals it can use to judge importance. For each rival, trace the path to a sample listing and write the steps down next to your own.

The result is not a call to copy the shallower tree. It is a prompt to check that each step has a meaningful label and a working link. Google tells you to use anchor elements with an href, and notes that Googlebot might not detect navigation driven by JavaScript events. Looking at a rival's page source for href attributes on menu items shows whether its menu follows that rule.

Sourcing terms without copying

A term list holds short names: category labels, facet labels and URL words. Two limits keep it fair. First, use the list for comparison and do not transplant a rival's whole tree into your site. Second, do not copy category intro text, listing descriptions or images. Write each category's description from your own stock knowledge. Practical suggestion: attach a note to the list saying that no rival text was taken in. Anyone who opens the sheet later then knows what it is and what it is not. The same note can name the date of each reading and the person who did it, which also settles questions about whether a label was current when it was recorded. If a rival later changes its menu, a dated row shows whether your decision was made before or after the change, and that context matters when a colleague asks why your tree differs from the rival's. It also protects the team from the temptation to treat an old reading as a current one.

Example scenario: a three-week vocabulary sprint

Example scenario: a used books marketplace wants to rename its menu. The team runs the work in three short stages.

Only the first stage depends on rival pages. The second and third rest on the team's own data and on the URL rules in Google's ecommerce documentation. A rival menu is a prompt for a final label, not the reason for it. For the structure side, see site architecture and breadcrumbs.

A worked pass: from label sheet to decisions

A label sheet becomes useful once each row ends in a decision. Take the sheet row by row with a fixed order of questions and write the answer in the same row, so that the reasoning stays attached to the term.

1. Does the label name a group of listings, or an attribute of a listing? A group is a category candidate and an attribute is a filter candidate.
2. Do you hold listings that fit it? If not, mark the row "hold" and revisit when stock arrives.
3. Does it duplicate a label already in your tree? If so, record it as a synonym of the existing node.
4. Does it fit under a parent already in your tree? If so, place it there; otherwise decide whether a new parent is justified.
5. Write the URL word and check that the menu, sitemap and canonical tag will carry the same address.

Example scenario: the decision table reads like a small meeting record. The team did not need rival data for any of the decisions in the last column, only for the first two columns. That split is the healthy one: rivals supply the vocabulary, your catalog supplies the answers.

Keeping the vocabulary alive

A vocabulary decays when new stock arrives and nobody updates it. Give the sheet owners.

Practical suggestion: reread rival menus at the start of each season and after a major catalog change, and update only the changed rows. Google's guidance can change, so check the current pages before you act.

The rejected labels column deserves care. When a colleague proposes "Pre-Owned Books" three months later, the sheet shows why "Used Books" won and saves a repeat argument. It also prevents two near-identical names from splitting one group of listings across two categories.

Checklist

• Rival menus and facets were opened by hand on public pages.
• Levels, labels, URL words and sources are in separate columns.
• Different names for one concept sit on one row.
• Each candidate was checked against your own listings.
• No category is opened without stock.
• One URL form is set for menu, sitemap and canonical tag.
• Spelling and plural variants are recorded in their own columns.
• No rival text, listing or image entered the sheet.
• The reading date is recorded.

Mistakes to avoid

Importing a rival's tree whole

A rival's tree fits the rival's stock. Categories for books you do not sell become empty pages.

Splitting synonyms into separate categories

Two categories for the same group repeat listings and divide the internal links that point to them. Keep one category and record the other name as a synonym.

Renaming without updating the URL

If the visible label changes while the URL word and sitemap stay the same, three places disagree. Follow Google's advice on one URL everywhere.

Reading the vocabulary as a ranking promise

A well-chosen label makes navigation clearer. Google's documentation makes no promise about rankings.

Bulk collection

Reading a menu by hand is public review. Automated bulk collection can break a rival's terms of use.

A decision rule for the term list

Before a rival label enters your tree, ask three questions of it. Does it name something you actually sell? Does a customer use that word when describing the product? Can you place it in one spot in your own menu without a duplicate? If any answer is no, leave the label on the comparison sheet and out of the live menu. A label that passes all three is written into the vocabulary with your own wording, a date and the reason for the choice.

Example scenario: a rival calls a group "Rare Collectibles" while your store sells only paperback classics and textbooks. The label fails the first question, so it stays on the sheet as a note about how the rival groups its stock, and your own tree keeps the narrower names.

Frequently Asked Questions

How do I read a rival's category names?

Open the public site, walk through the menu level by level, and record each label, its breadcrumb, its page heading and the URL words. Add the facet groups from a category page on a separate sheet.

What does Google say about category names and URLs?

Google says it analyzes the linkages between pages to understand site structure and recommends descriptive anchor text and descriptive words in URL paths. It also advises using the same URL in internal links, sitemaps and canonical tags.

Can I turn all extracted terms into categories?

No. A term becomes a category only if you hold listings for it and your own search data supports the name. A rival's use of a term is not enough on its own.

Should facet names appear in URLs?

Google recommends ?key=value parameters for things like filters. Whether filtered pages should be indexable is a separate decision handled in the faceted navigation guidance.

What should I do with a category that has no items?

Google suggests a noindex robots meta tag for a category with no items. If the site removes empty categories from search and browse automatically, it suggests considering a 404 status code.

Is copying a rival's category descriptions acceptable?

No. Keep the term list to short names and write each description from your own stock knowledge. Do not copy intro text, listing descriptions or images.

Conclusion

Mining a rival's category and filter names gives your own naming a factual starting point. Read the menus and facets by hand, record labels and URL words, map synonyms, test each candidate against your stock and search data, and place the winners in your tree with one URL form throughout. The vocabulary supports your decision. It does not replace it.

When your category and listing pages are ready, you can also build links toward them. Explore the backlink packages to see the options.