Ecommerce SEO
Make a large catalogue findable without letting it crawl out of control
Ecommerce SEO is mostly an architecture problem. Facets multiply URLs, collections compete with categories, variants split one product across a dozen thin pages, and stock changes daily. We decide what should exist, what should be indexed, and what should redirect.
What a catalogue does to a site
Your store generates URLs faster than any crawler will get through them
A brochure site has forty pages and a person decided on every one of them. A store has a template, a taxonomy and a set of filters, and between them they produce URLs nobody chose.
That is the real subject of ecommerce SEO. Not writing better product copy — deciding what should exist at all.
- Filters have turned one category into forty thousand addresses.
- Colour, size, price band, brand, rating and sort order combine into permutations that no merchandiser ever intended to publish. Each is a real URL that can be linked, crawled and sometimes indexed. The damage is rarely duplicate content. It is that crawling gets spread across permutations while the fifty pages carrying the revenue are revisited slowly.
- The collection page and the category page are fighting each other.
- Merchandising built categories. A campaign built collections. Both now target the same commercial search, both have some links, neither has enough. Search engines pick one, often the weaker one, and the internal linking keeps sending signals to both. Consolidating them properly is usually worth more than any amount of new content.
- Every variant is its own thin page.
- Twelve colours become twelve URLs with the same forty words and a different swatch. None of them individually deserves to rank, and together they dilute the one page that would. The fix is a decision about which variant is canonical and how the others are exposed, and it is a template change rather than a content project.
- The product description is the manufacturer feed.
- If your page says exactly what nineteen other retailers say, there is no reason for a search engine to prefer yours and no reason for a shopper to buy from you rather than the cheapest listing. The differentiator is the information only a seller who handles the product can add: fitment, real photography, sizing that reflects returns data, compatibility, and what it is not good for.
- Discontinued products are deleted.
- A product that sold for three years has links, internal signals and accumulated relevance. When the SKU is retired and the row is removed from the database, all of that goes with it. Over a few years of ordinary catalogue churn this quietly removes a significant share of a site accumulated authority, and nobody ever attributes the decline to it.
Two different jobs
Category pages and product pages are not competing for the same searcher
Most stores put their writing effort into product pages and their navigation effort into categories. The commercial demand usually sits the other way round.
| Dimension | Category and collection pages | Product detail pages |
|---|---|---|
| What the searcher is doing | Still choosing. They know the type of thing they want but not which one. | Already chosen, or checking price and availability for a specific item. |
| Typical search | A category plus a qualifier: material, use case, size range, price band, audience. | A model number, a brand plus product name, or a barcode. |
| What the page has to prove | That you have genuine range and can help someone narrow it down — filters that make sense, buying guidance, honest comparison. | That this specific item is the right one, in stock, and worth buying from you rather than elsewhere. |
| Who you are competing with | Other retailers, editorial round-ups and marketplaces with deeper range. | Every other reseller of the identical item, including the manufacturer and the marketplaces. |
| Most common failure | A grid of products with a heading and no content, or three near-identical pages splitting the same demand. | Manufacturer copy, no unique media, variants split across a dozen URLs. |
| Where the durable value is | High. Category demand persists after individual products are discontinued. | Lower per page, high in aggregate, and it evaporates when the SKU is retired. |
How we work through it
Decide the catalogue architecture before writing a word
The order is deliberate. Content written before the URL decisions are made usually has to be moved, merged or thrown away.
Inventory the URL space
A full crawl plus log files and Search Console coverage data, so we can see the difference between what the navigation exposes, what gets crawled and what actually earns anything. On most stores this is the first time anybody has seen the true size of the site.
You get: URL inventory segmented by template and catalogue level
Set the facet policy
For each attribute, a decision: indexable with real content, crawlable but not indexable, or blocked entirely. The decision is made against evidence of demand for that attribute combination, not a blanket rule copied from a blog post.
You get: Written facet policy per attribute, with the implementation method for your platform
Consolidate the category layer
Identify every place where two or more pages target one intent, choose the survivor on links, history and fit, and map the redirects. This is where the largest single gains usually are on a store that has been trading for a few years.
You get: Consolidation and redirect map, ready for the release
Fix the templates, not the pages
Variant canonicals, structured data, breadcrumb logic, pagination behaviour, internal links from category to subcategory to best sellers. Template changes apply to the whole catalogue at once, which is the only way anything scales past a few thousand SKUs.
You get: Template specifications the development team can build from
Tier the content effort
Write properly for the categories and hero products that carry the demand. Use a structured content model for the middle of the catalogue so every page has genuine attributes and comparisons. Accept that the deep tail will not be a search asset and stop spending on it.
You get: Content tiering plan and written pages for the top tier
Set the lifecycle rules
What happens when a product goes out of stock, gets discontinued, or comes back next season. Written down, agreed with merchandising, and ideally automated in the platform so it does not depend on somebody remembering.
You get: Product lifecycle policy and redirect rules
Scope
What the work covers
Scope is set against the size and state of your catalogue. A two hundred product store and a two hundred thousand product store need different work under the same service name.
- Crawl, log file and index coverage analysis across the full URL space
- Faceted navigation audit and a written per-attribute indexing policy
- Category, subcategory and collection consolidation with a full redirect map
- Category page content model — what each page must say beyond listing products
- Product template review: canonicals, variants, media, specifications and reviews
- Product, Offer and BreadcrumbList structured data implementation and validation
- Out-of-stock, discontinued and pre-order handling rules
- Seasonal range strategy using durable URLs rather than annual rebuilds
- Pagination and infinite scroll behaviour that does not hide products from crawlers
- Internal linking from categories to subcategories, best sellers and buying guides
- Sitemap segmentation so indexing can be measured per catalogue area
- Site search and zero-results analysis, which is the cheapest demand research a store owns
How shoppers actually search
The searches a store has to cover
These are illustrative patterns rather than a keyword list. The point is that they land in different places in the catalogue, and a store that only covers one of these patterns is leaving the others to competitors.
Category-level buying searches
These are the highest-value searches most stores are worst at. They need a category page with a point of view, not a product grid.
- waterproof hiking boots wide fit
- commercial coffee grinder for a small cafe
- best office chair for lower back pain
- stainless steel garden furniture uk
Product and model searches
You are competing with every other reseller of the identical item. What you add beyond the feed is the entire argument.
- brand name model 3200 review
- part number cross reference chart
- model x replacement filter compatible
- buy specific product name in stock
Pre-purchase questions and comparisons
These pull people into the catalogue earlier and are usually the only searches where a retailer can outrank a marketplace.
- difference between ceramic and cast iron pans
- what size mattress fits a standard uk double bed
- is this compatible with my machine
- how long does delivery take on furniture
These are examples of how customers in this market search, drawn from keyword research and from the questions that come up on sales calls. They are illustrative, not a volume claim — the actual demand in your area is something we size before recommending anything.
Questions
What ecommerce teams ask us first
Should faceted navigation URLs be indexed?
Some of them, and only where the combination matches a search people actually make. A facet like "womens running shoes size 8" may deserve an indexable page in a large catalogue. A facet combining four attributes and a price band almost never does.
The decision is made per facet, not per site. We map which attribute combinations have real demand, allow those to be crawled and indexed with proper titles and copy, and close the rest off so they neither absorb crawling nor compete with the category above them.
What should happen to a product page when the product is out of stock?
It depends on whether the product is coming back. If it is temporarily unavailable, keep the page live, mark the availability accurately in the structured data, and give the visitor something useful to do — a restock notification, a size that is in stock, or the nearest alternative.
If the product is gone for good, the page should redirect to the closest equivalent product or to the parent category. Deleting it and serving a 404 throws away every link and every internal signal it had.
Do we need unique descriptions for 40,000 products?
No, and trying is usually a waste of budget. The realistic approach is to tier the catalogue: write properly for the products that carry the revenue and the search demand, use a structured content model for the long tail so each page still has genuine attributes, comparisons and specifications, and accept that the deepest tail will never be a search asset.
What matters more than word count is whether a page says anything the other twenty resellers of the same product do not. Fitment notes, real photography, sizing guidance, compatibility and honest limitations do that. A rewritten manufacturer paragraph does not.
Category page or collection page — which should we use?
One of them, not both. Most stores end up with both because the taxonomy was built once by the merchandising team and again by whoever ran a campaign, and the two now split the same intent between them.
We look at which has the links, the history and the better fit for how people search, make that the canonical destination for the cluster, and redirect the other. The consolidation is usually one of the highest-value changes available on a mature store.
How do you handle seasonal ranges?
By keeping one durable URL per recurring range rather than creating a new one each year. A page at the same address every year accumulates links and relevance; a new page each season starts from nothing every time.
Out of season, the page stays live and explains what is coming, links to the current alternative, and holds its position. In season, it becomes the landing page again.
Will this work on Shopify, or do we need to replatform?
Almost always yes, and almost never. Every major platform has constraints — fixed URL prefixes, limited control over facet parameters, template restrictions on how variants are exposed — and every one of them can be worked with.
We will tell you honestly if a constraint is genuinely costing you money rather than just being untidy. Replatforming for SEO reasons alone is rarely the right call, and a migration done badly costs more than the problem it was meant to fix.
Find out how big your catalogue actually is
Before a strategy call we crawl the store and count what is really there — the facet permutations, the competing category pages, the orphaned products. The number is usually a surprise, and it tells you immediately where the work should start.
If we don't deliver the work we agreed to deliver for reasons within our control, you don't pay for the undelivered work. Read our guarantee
Related
Where to go next
- the technical side of a large catalogueCrawling, indexing and rendering decisions sit underneath everything here.
- how we work with online retailers
- growing online salesThe commercial outcome this service serves.
- shopping and Performance Max campaignsThe same feed and the same page quality problems apply.
- the wider organic programme
Last updated · Reviewed by Zubair Afzal