Most of this genre is general advice with the word ecommerce inserted: write good titles, get links, be fast. All true, none of it specific. So this ecommerce SEO guide skips them and starts with the two problems that genuinely belong to selling online — they are where most of the opportunity sits, and almost nobody works on them because neither is visible from the front of the site.
Problem one: your platform generates duplicate URLs by itself
A catalogue does not have one URL per product. It has one canonical URL per product and then a long tail of variations the platform creates on its own: the same product reachable through every collection it belongs to, filter and sort parameters, pagination, search result pages, tag archives, and variant parameters appended to the address.
Each of those is a page as far as a crawler is concerned. On a catalogue of any size that is thousands of near-identical URLs competing with the ones you actually want ranked, and spending the crawl budget that should be going to your real pages.
The fix is unglamorous and it is the highest-value work available on most stores. Every product should declare one canonical address and stick to it regardless of the path the visitor took. Filtered and sorted views should not be indexable — they are navigation, not content. Pagination should be crawlable so deep products are reachable, without each page pretending to be a destination. Internal search results should never be indexed at all.
None of that changes anything a customer sees, which is exactly why it never gets prioritised and why it is still available on nearly every store we look at.
Problem two: the collection page is the money page
Stores pour effort into product pages. Product pages mostly capture people who already know what they want, often by brand and model, and that is a small and late slice of demand.
The larger, earlier demand is categorical: someone wants a kind of thing, not a specific one. That query is served by the collection page, and on most stores the collection page is a heading, a grid, and nothing else — no reason for it to rank ahead of the twenty other shops with the same grid.
A collection page that earns the position answers the questions a buyer has before they can choose: how the options differ, what the sizing or compatibility trap is, what to buy for which use. That is a few hundred words of genuinely useful writing, placed where it does not push the products below the fold. It is the single biggest structural opportunity on most catalogues, and it compounds, because one collection page can serve dozens of variations of the same query.
Product pages: the parts that actually matter
Manufacturer descriptions are on every retailer's site selling the same item, which means your page has nothing of its own. Rewriting them is dull work and it is the difference between a page and a duplicate.
Structured data should describe the product honestly — price, availability, reviews if they are real. The temptation to mark up ratings you do not have is a short road to a manual penalty, and it is the most common way stores get themselves into trouble.
Reviews are the cheapest genuine content you will ever get, and they update themselves. A product with forty reviews has forty pieces of language written by people describing the item the way other buyers would search for it, which is a vocabulary no copywriter reliably guesses.
And handle out-of-stock properly. Deleting the page throws away everything it earned; leaving it live with no alternative wastes the visit. Keep it, say so, and offer the nearest thing.
Speed, in the only terms that matter
Speed is a ranking factor and a much larger conversion factor, and on most stores the causes are the same two every time: too many apps loading on every page, and images uploaded at full resolution.
Measure with the field data in Search Console rather than a synthetic score out of a hundred. The lab tool tells you how a page performs on a machine in a data centre; the field data tells you how it performed for your actual customers on their actual phones, which is the only version with money attached.
What to do first, in order
Fix the indexing problems before writing anything. Publishing new pages onto a site that is already competing with itself is pouring water into a leaking bucket, and the fixes are quick.
Then pick your five highest-intent collection pages and write them properly. Not all of them — five. If that does not move anything in three months, the diagnosis was wrong and it is better to learn that from five pages than from fifty.
Then product pages, starting with the ones that already get impressions but few clicks, because those are pages Google is willing to show that people are not choosing. That is a title and description problem far more often than a ranking problem, and it is the fastest win left on most stores.
Only then think about links. Not because links do not matter — they matter a great deal — but because links to a site that is competing with itself, on collection pages with nothing on them, are an expensive way to make a structural problem more visible.
If you want to know which of these your store is actually losing to, that is what the free written teardown is for; it comes back with the list in priority order. The ongoing version of this work is described on the SEO page.
