Category Page SEO and Optimization: How to Build PLPs That Rank and Answer Shopper Intent

Broad queries are category queries

Almost nobody starts with a SKU. People type "a linen dress for a summer wedding", and classic search answers with category pages, the only page type that matches a set of products, a use and a price range at once. Ahrefs' ecommerce SEO guide puts category and product pages at the center of ecommerce SEO because they "tend to be the most lucrative" pages on a store.

Check it in your own data. In Search Console's Performance report, filter out queries containing your brand name and open the Pages tab. On most catalogs we look at, category URLs sit near the top.

AI agents fan out into category questions

Google describes AI Mode running several searches at once to work out what makes a product right for a specific need. In its own example, AI Mode figures out "what makes a bag good for rainy weather and long journeys" before it shows any products (Google). Each sub-query is a category question: waterproof travel bags, carry-on backpacks, bags with easy-access pockets. A clear page per set is far easier for an agent to read than one "Bags" page with 400 products. More on the mechanism in Google AI Mode shopping.

Category pages frame the comparison

When an agent compares options, it looks for the criteria that matter in that category. Fabric weight for linen. Drop and cushioning for running shoes. Noise level in decibels for dishwashers. A good PLP states those criteria in plain text, and agents reuse that framing when they explain a pick. That's why we treat category pages as part of Agentic Commerce Optimization (ACO): the same product data feeds the PLP, the PDP, the Google Merchant Center feed and every agent that reads them.

Product pages win the specific match; category pages win the question that comes before it.

Much SEO advice treats category copy as keyword text parked at the page bottom. For an agent, it's where your store explains how to choose.

Structure and placement

Working ranges only. A narrow PLP needs less, a broad hub like "Running Shoes" more.

One popular tip we'd skip: 300+ words of SEO text above the grid. On mobile it pushes the first product row off screen, and a guide below the grid does the same job better.

What makes category copy useful

• Selection criteria. Name the attributes that separate products in the category, with values: "Lightweight linen (around 150 gsm) for heat, mid-weight (180 to 200 gsm) for structure and less wrinkling."
• Use cases and occasions. Weddings, office, travel, beach. That's how people phrase AI prompts.
• Comparisons. "Midi vs maxi". Agents look for exactly this kind of contrast when they justify a pick.
• Shared vocabulary. Use the terms shoppers search for, and the same attribute names your PDPs and feed use, such as material, color, size, pattern and age_group.

Example: a category description, before and after

A linen dresses collection from a fictional brand, Saltmarsh:

The "before" copy could sit on any store. The rewrite names material, size range, lengths, styles and occasions. A shopper sees where to click, and an agent fanning out "linen dress for a summer wedding, petite" can confirm the store covers every part of the request.

Writing the category FAQ

Category FAQ questions come from "People also ask" for the category keyword, from customer service tickets and return reasons ("runs small" is an FAQ waiting to be written), and from the prompts shoppers type into AI assistants. For linen dresses:

• Does linen wrinkle, and how do I reduce it?
• Is a linen dress appropriate for a wedding?
• What is the difference between 100% linen and a linen blend?
• How should a linen dress fit if I am between sizes?
• Can I machine wash linen dresses?

Answer each in two or three plain sentences, consistent with your PDPs. An agent that reads "machine washable" on the PLP and "dry clean only" on the product will trust neither. For product-level writing, see How to write product descriptions.

If you sell in several markets, localized PLPs need their own demand research: a translated intro misses the local head term, and Italian pages quoting US sizes 0 to 22 are a regular find. See Ecommerce localization.

Why facets are a problem

Baymard found that 51% of sites don't offer all five essential filter types (price, average user rating, color, size and brand), and 14% don't let users select more than one value in a filter type (Baymard Institute). Earlier Baymard testing found that sites with mediocre product list usability saw abandonment rates of 67% to 90%, against 17% to 33% on sites with even a slightly optimized toolset (Baymard Institute).

Those same filters create the crawling problem. A few filter types combined freely, times sort orders and pagination, produce huge numbers of near-identical URLs. Google says crawling them uses server resources and can slow the discovery of your new pages (Google Search Central).

Settings > Crawl stats in Search Console lists example crawled URLs; with uncontrolled facets it fills up with parameter strings (Magento's layered navigation produces URLs like ?color=49&material=112 from attribute option IDs).

Decide which facets deserve to be indexed

A facet earns an indexable URL when people search for it, it holds enough products, and you can write something specific about it.

The opposite mistake is just as common: noindexing all filters by reflex and giving away "linen dresses" or "petite dresses", often the highest-demand pages in the category tree. A short, written allow list, reviewed quarterly, works better; our guide to faceted navigation SEO and indexable facets shows how to build one.

Google's rules for faceted URLs

• Block what you don't need indexed

  Google suggests robots.txt rules for filter parameters, such as /*?*color= or /*?*size=, while keeping an unfiltered listing crawlable.

• Or use URL fragments

  Filters placed after a # are ignored by Google's crawling and indexing, so those URLs never add to crawl load.

• Use the & separator

  Google asks for the industry-standard & between parameters; commas, semicolons and brackets are often not read as separators.

• Keep filter order consistent

  In path-based filters, always output the same order and prevent duplicate filters in one URL.

• Return 404 for empty results

  When a combination returns no products, or is a duplicate or nonsensical combination, return an HTTP 404 instead of redirecting.

• Treat canonicals and nofollow as weak signals

  Canonicals to the unfiltered page reduce crawl only over time; nofollow works only if every link to that URL carries it.

Source for all six rules: Google Search Central, managing crawling of faceted navigation URLs.

"Canonical every filter URL to the parent" is still the default fix in many audits. Fine as a safety net. As the main control on a large catalog, it leaves Googlebot fetching URLs it will mostly ignore.

Turning high-demand facets into real landing pages

A filter URL like /dresses?material=linen&length=midi makes a weak landing page: generic H1, no copy, parameters that may be blocked. When the data shows demand for "linen midi dresses", promote it to a static PLP:

1. Give it a clean, permanent URL, such as /dresses/linen-midi-dresses.
2. Write an H1, intro, buying guidance and FAQ specific to linen midi dresses.
3. Define the product set with rules (material = linen, length = midi) so it updates itself.
4. Link to it from the parent category, from sibling PLPs and, where your platform allows, from the filter itself.
5. Keep the parameter version out of the index so the two URLs don't compete.

If a URL is blocked in robots.txt, Google can't read a noindex tag on it. Choose one mechanism per URL pattern: block crawling for patterns that should never be fetched, or allow crawling with noindex for pages you want dropped from the index first.

The best faceted navigation strategy indexes a short list of facets on purpose and keeps everything else out of the crawl by default.

The fields that matter

Shopify confirms that the H1 on product and collection pages comes from the Title field, and gives the title and meta description guidance above (Shopify Help Center).

Automated collections for demand-based PLPs

Shopify's automated (smart) collections add products automatically based on conditions, with up to 60 selection conditions and a choice between matching all or any of them (Shopify Help Center). That makes them the natural tool for demand-led PLPs: "Linen midi dresses" becomes product type = Dress AND metafield material = Linen AND tag = midi. Shopify notes that a new collections model is replacing the legacy manual and smart collections model, so check which version your admin shows before you build at scale.

Prefer metafields to tags for these rules. Tags drift ("midi", "Midi", "midi-length"), and one typo silently drops a product.

Filters and metafields

The Shopify Search & Discovery app powers storefront filters. Standard filters cover availability, category, price, product type, tags and vendor. Custom filters can use product options (like Size), product metafields, category metafields, variant metafields and standard product attributes from Shopify's Standard Product Taxonomy. A store can have up to 25 filters, a filter shows at most 100 values on the storefront, and collections with more than 5,000 products don't display filters at all (Shopify Help Center).

Split very large collections into narrower, demand-led ones, and keep attributes in metafields so they drive filters and collection rules alike. Then check those values reach your Google Merchant Center feed; custom metafields often need explicit mapping in the channel app.

Crawling and duplicates on Shopify

• robots.txt. Shopify's default robots.txt already blocks admin, cart, checkout and some filtered or sorted collection URLs. Shopify calls editing robots.txt.liquid an unsupported customization that can cause loss of all traffic if done incorrectly (Shopify Help Center). Leave the defaults alone unless you have a specific, tested reason.
• Product URLs inside collections. Many themes link products as /collections/linen-dresses/products/aurelia-midi. Shopify canonicalizes these to /products/aurelia-midi, but linking straight to the canonical product URL keeps internal links clean.
• Tag URLs. Tag pages like /collections/dresses/linen reuse the parent collection's copy. If "linen dresses" deserves a page, make it a collection.
• Thin collections. Seasonal or campaign collections with three products and no copy should be merged, hidden from navigation or set to noindex once the campaign ends.

Re-check impressions and clicks per collection in Search Console four to eight weeks after each change. Beyond collections, see Ecommerce SEO.

How the platform maps to the work

Why the product data comes first

A category page can only be as precise as the attributes behind it. That's why, in most projects, we fix the products before we build new PLPs. The Optimizer extracts attributes such as material, fit and use case and writes them back as structured data, which rule-based PLPs and filters need. The same data generates Google Merchant Center conversational attributes for your Shopping feed.

Results from brands and retailers using AndromedAI

AndromedAI has optimized more than 500 catalogs. Some results from brands and retailers on the platform:

Start with the free AI Readiness Audit: send three product page URLs and see what AI agents and search engines can and can't read.