Ecommerce SEO is an architecture problem before it is a content problem
A brochure site has forty pages. A store has forty thousand URLs, most of which it created by accident, and the job is deciding which ones deserve to exist.
The same three faults show up in almost every store we audit. Faceted navigation generating a crawlable URL for every combination of colour, size and price band. Category pages competing with each other because someone created both “running shoes” and “shoes for running”. Product descriptions copied from the manufacturer feed, identical to the two hundred other retailers selling the same item.
None of those are fixed by publishing blog posts. They are fixed by deciding what the site’s URL structure is supposed to be, and then enforcing it.
What we do on a store
Faceted navigation and crawl budget
Decide which facets are demand (indexable, linked, given real copy) and which are filters (noindexed or blocked). Then enforce it in robots rules, canonicals and internal links so crawlers spend their budget on pages that can sell.
Category page architecture
One category per genuine intent, mapped against search demand rather than your internal merchandising tree. Copy above or below the grid that answers the buying question, and an internal link structure that pushes authority to the categories with margin.
Product pages at scale
Templated but not duplicated: unique intro copy, real specifications, structured data that matches the price and stock actually shown, review markup only where reviews genuinely exist, and a plan for variants that does not create nine URLs per product.
Out-of-stock and seasonal handling
What happens to a URL when the product dies is one of the largest silent traffic leaks in ecommerce. We set the rules: keep and re-merchandise, redirect to the parent, or let it 410 — and we make the platform follow them.
Platform playbooks
Most ecommerce SEO problems are configuration, not code. These are the ones specific to each platform.
Shopify is fast to launch and opinionated about URLs, which causes a predictable set of problems.
- Duplicate product URLs under /collections/*/products/* competing with the canonical /products/*
- Collection pagination and sort parameters generating crawlable duplicates
- Theme-level Core Web Vitals damage from app scripts nobody audits
- Blog structure stuck under /blogs/news/ with no topical grouping
- Metafield-driven structured data, so schema stays true as stock changes
WordPress gives you full control, which means full responsibility for what gets indexed.
- Attribute and tag archives generating thousands of thin pages by default
- Product category depth creating five-level URLs nobody can rank
- Plugin conflicts producing two canonical tags on the same page
- Server-side caching that breaks cart behaviour and gets disabled site-wide as a result
- Image handling — the single biggest Core Web Vitals win on most Woo stores
Enterprise stacks need SEO expressed as tickets and acceptance criteria, not as advice.
- Layered navigation configuration and its crawl consequences
- Multi-store and multi-currency URL strategy, including hreflang
- Server-side rendering checks on JavaScript-heavy category grids
- Release-cycle involvement, so SEO requirements land in the sprint rather than after it
If most of your revenue is on Amazon or Flipkart, that is a different discipline with a different algorithm, and we will say so rather than sell you site SEO.
What we do handle is the relationship between the two: making sure your own store ranks for the brand and category terms marketplaces will never give you, so you are not renting your entire demand.
What we look at first on a store
These four numbers usually tell us where the money is going before we read a single page of content.
80%
of crawl budget wasted on parameter URLs
Typical of stores we audit before facet control
4
clicks to reach a money category
Should be two. Depth costs rankings
0
unique words on a PDP
Manufacturer feed copy, shared with every competitor
12%
of SKUs generating all the organic revenue
Where the content budget should go first
Where content actually earns on a store
Not a blog. Four specific page types, in this priority order.
First
Category copy that answers the buying question
A buyer landing on a category is choosing between options, not learning about the product type. Two hundred words that help them choose — sizing, materials, what to avoid — outperform a thousand words of keyword-led preamble.
Second
Buying guides tied to the category
One guide per genuine decision, linked from the category and linking back to it. These capture the research query and pass authority to the page that sells.
Third
Comparison and alternative pages
High intent, low volume, and almost nobody builds them for physical products. “X vs Y”, “best X for Y use case” — written honestly enough to be trusted.
Fourth
Support content that prevents returns
Sizing, care, compatibility, installation. Ranks well, reduces returns, and earns links from forums and communities that no outreach campaign could reach.
Ecommerce SEO FAQs
We have 20,000 products. Do you write copy for all of them?
No, and you should not pay anyone to. We find the SKUs and categories that carry the margin and the demand — usually a small fraction of the catalogue — and invest the writing there. The rest get templated structure and correct markup.
Will fixing faceted navigation lose us traffic?
Some indexed parameter URLs do receive traffic, which is why we model it before touching anything: which facet URLs earn, which are noise, and what the redirect plan is. Doing this blind is how agencies cause the drop they were hired to prevent.
Do you handle Amazon listings?
Marketplace optimisation is a separate discipline and we do not claim it. We will help you make sure your own store owns the demand a marketplace cannot take from you.
How long before organic revenue moves?
Technical fixes on a store can move revenue faster than on a brochure site, because you are unblocking pages that already have buying intent — sometimes inside two months. Category content and authority work stay on the usual four-to-nine-month curve.