About4 min read
Ecommerce SEO Is Mostly An Engineering Job
General SEO advice assumes every page is deliberate. A store generates thousands automatically from a catalog that changes daily, and that difference produces the problems that actually decide whether it ranks.
By Juan Molina
On this page
Ecommerce SEO is not general SEO applied to products. The advice that works for a blog assumes every page is a deliberate piece of content, and a store generates pages automatically, by the thousand, from a catalog that changes daily.
That difference produces a specific set of problems, and they are the ones that decide whether a store ranks. None of them are about keyword density.
The structural problems that actually matter
Faceted navigation, which can generate infinite pages
Filters are the single biggest technical SEO problem in ecommerce. Colour, size, price, brand, sorted three ways, and every combination is a URL. A catalog with six filters can produce more crawlable URLs than you have products, most of them near-identical.
Search engines waste their crawl budget on that, and your actual category pages compete against thousands of variations of themselves.
The decision to make, per filter: does this combination deserve to rank? A page for blue running shoes probably does, because people search it. A page for blue running shoes sorted by price ascending does not. The first should be a real indexable URL with its own title and copy. The second should be blocked, canonicalised or not linked at all.
Make the list explicitly rather than letting the default behaviour decide.
Variants competing with each other
The same shirt in eight colours as eight URLs with nearly identical text is eight pages splitting the authority that should belong to one.
Usually the right shape is one product page with selectable variants and a canonical pointing at it. The exception is where people genuinely search for the variant, which happens with things like sizes and capacities, and there the variant page earns its own existence and its own content.
Category pages with nothing on them
Category pages are the ones with real commercial search volume, and most are a heading and a grid. Nothing to rank on.
The fix is not a wall of text underneath the products, which nobody reads and which fools nobody. It is a short piece of genuinely useful buying guidance near the top, the questions people ask about that category answered, and internal links to the subcategories that matter.
Products that go out of stock
Deleting the page throws away everything it earned. Redirecting it to a category loses the specific term. Leaving it live with no way to buy is a bad experience.
Keep the page, keep it indexed, say clearly that it is unavailable, and offer alternatives on the same page. If it is gone permanently, redirect to the nearest equivalent product rather than to the category. And ensure your structured data says out of stock rather than lying, because the mismatch gets noticed.
The rendering problem
If your product data only appears after JavaScript runs, you have a different and larger problem, and it is worth reading about separately. Check by viewing source on a product page and searching for the price.
What to do in what order
Roughly, and it is worth being disciplined about the order, because most stores start at the bottom.
One. Confirm your pages render server-side. Nothing else matters if they do not.
Two. Fix faceted navigation. Decide which filter combinations are indexable and block the rest.
Three. Sort out variant canonicals so one page holds the authority.
Four. Give your top twenty category pages real content. These have the volume.
Five. Product schema on every product, accurate about price and availability.
Six. Then, and only then, think about a blog.
Most stores do six first because it is the most enjoyable, and publish content that points at a technically broken catalog.
The keywords that are worth your time
Category terms are where the money is. A page that ranks for a category term serves every product on it, which is a much better return than a product page ranking for one item.
Long-tail product terms convert best and are usually uncontested, because they are specific enough that nobody optimised for them.
Comparison and alternative terms are where people are deciding, and they are usually easier than the head term.
Head terms like the name of your whole product category are dominated by marketplaces and retailers with enormous authority. Know that going in rather than spending a year discovering it.
Where we fit, and where we do not
The parts a backend controls are the technical ones. Because the storefront is yours, URLs, canonicals, filter behaviour and rendering are decisions you make rather than platform conventions you work around, which is exactly the set of things that go wrong on hosted platforms. Product data including stock is available over the API, so structured data can be accurate rather than a snapshot.
Where we are the wrong answer: that control cuts both ways, because you now own those decisions and can get them wrong on your own rather than inheriting someone else's reasonable defaults. If nobody on your side is going to think about canonicals, a hosted platform's defaults are better than your neglect. And no backend writes category copy, which is the part with the most upside.
The short version
Check rendering. Control your filters. Canonicalise variants. Put real content on category pages. Keep out-of-stock products alive. Blog last.
Four of those six are structural decisions rather than content, which is why ecommerce SEO is mostly an engineering exercise that people try to solve by writing.
