Ecommerce SEO: the complete guide for online stores

Stores rarely lose organic visibility in one single place. Usually structure, page types, technical setup and content are each half right, and the result is mediocrity spread across thousands of URLs. This guide works through the six areas in the order that pays off fastest: architecture, category pages, product pages, technical foundation, content and internal links, and finally measurement. Every section ends with concrete steps.

1. Architecture: the map before the work

Your page structure decides which searches you can serve at all. It comes before any copy, because good copy on a page that does not exist helps nobody.

Page types and their job

The most common gap sits between subcategory and product page: for many in-demand property combinations there is either no indexable page at all, or only a filter URL that technically cannot rank.

A second, less noticed gap is brand pages. If you resell other brands, people search for brand plus product group. One page per brand that shows that brand's range and briefly explains what the brand stands for serves those queries. It behaves like a category, and in many stores it is either never created or exists only as a logo list without its own URL.

Rules for the build

• Keep it flat. Every important page should be reachable in at most three clicks from the home page.
• One intent, one page. Two pages on the same topic weaken each other.
• Name categories after demand, not after internal jargon. If customers search for rain jacket and you write weather protection outerwear, you lose.
• Treat navigation as the blueprint. What does not appear in the main nav or footer is harder for Google to find and harder for people too.
• Short, stable URLs. Readable, lowercase, no IDs or parameters, and not changed after launch.

Steps for this area

1. Export every indexable URL in your store and sort them by page type.
2. For each top category, note the three to ten subcategories that should exist.
3. Mark the missing pages and the topics that are covered twice.
4. Define a target depth and check which categories sit deeper than that.

2. Category pages: the revenue lever

Category pages serve the queries with volume and buying intent. At the same time they are the weakest pages in most stores, because they consist of nothing but a product grid.

What belongs on a category page:

• An H1 that matches the main query, in the language customers use.
• A short intro above or directly under the H1 saying what is here and how to choose. Two to four sentences; the products must stay visible.
• A sensible default sort order that puts available and popular products first.
• Filters customers actually use, wired up so they do not spawn an endless set of indexable URLs.
• Links to subcategories and related categories as visible elements, not only in the navigation.
• Optionally a longer section below the listing with buying criteria, care, sizing and common questions.

One frequent mistake: discontinued categories get deleted. If a category had clicks, redirect it to the closest successor category, not to the home page. A redirect to the home page is usually treated like a soft error rather than a real replacement.

3. Product pages: unambiguous, complete, persuasive

Product pages rank for specific searches and have to do two things at once: answer the user's question and differ from identical pages in other stores.

Non-negotiables

• Title and H1 with brand, model and the one decisive attribute.
• Your own description. Manufacturer copy sits word for word in hundreds of stores. If you paste it unchanged you have no reason to be preferred. If you use generative tools for the long tail, the guardrails are in the article on AI product descriptions.
• Structured specifications in table form: dimensions, material, compatibility, what is in the box. Those are often exactly the details people search for.
• Price, availability, delivery time and shipping cost visible on the page, not first in the cart.
• Images from several angles, with descriptive file names and alt text.
• Reviews, if you have real ones.
• Structured data of type Product with price and availability so Google reads the values reliably, see schema.org/Product and the walkthrough in product schema markup.

Variants and sold-out products

Variants such as size and colour are the most common source of duplicate content in stores. Decide deliberately: either one page per product with a variant selector, or separate pages per variant when people genuinely search for the variant. Doing both produces a mass of near-identical pages. On Shopify this has its own set of URL rules, which are covered in Shopify product variants and SEO.

For permanently sold-out items: keep the page and point to alternatives as long as it has demand. If a product is discontinued with no successor, redirect to the category. Rows of 404s cost you clicks you had already earned.

What makes product pages fail

Three patterns turn up in almost every store audit:

• Every product has the same layout with different numbers. If only dimensions and price differ, you create thousands of near-identical pages, and Google indexes only part of them.
• The copy does not answer the question that leads to a purchase. People want to know whether something fits, lasts, is compatible and when it arrives. Those details are often missing while adjectives are abundant.
• The page is only reachable through the category listing. With thousands of products that means effectively no internal linking for deeply buried items.

A workable prioritization: sort products by revenue over the last twelve months and rewrite the top ten percent by hand. For the long tail, complete and correct attributes plus one short original paragraph are enough.

4. Technical foundation

Technical work rarely wins rankings, but it reliably prevents them. Four areas decide the outcome.

Indexing

Google has to find the right pages and ignore the wrong ones. In the index: home page, categories, relevant filter pages, products, guides. Not in the index: cart, account, internal search results, sort and pagination parameters with no purpose of their own, and duplicates.

Important: robots.txt blocks crawling, not indexing, and a page blocked in robots.txt cannot have its noindex read. If a page should leave the index, it has to be crawlable and carry noindex.

The status you will deal with most in the page indexing report is "Crawled - currently not indexed", and on large stores it usually means quality or duplication rather than a bug. The diagnosis path is in crawled, currently not indexed.

Duplicates and canonicals

Every page needs a self-referencing canonical; duplicates point at the original. Typical sources in a store are variants, filter combinations, campaign parameters, products living in several categories, and print views. The article on duplicate content in ecommerce works through the cases one by one.

Speed

Core Web Vitals come with documented targets: LCP 2.5 seconds or less, INP 200 milliseconds or less, CLS 0.1 or less, each assessed at the 75th percentile of page loads and split between mobile and desktop. The usual culprits in stores are oversized images, too many third-party scripts and layout shifts from banners or review widgets loading late. The official definitions are on web.dev, the store-specific fixes in Core Web Vitals for online stores.

Crawlability and delivery

• A current sitemap containing only indexable URLs, submitted in Search Console.
• Everything on HTTPS, one single canonical domain variant, all others 301-redirected.
• Internal links pointing at the final URL, not into redirect chains.
• A complete mobile rendering, because Google evaluates the mobile version.
• Content present in the HTML without JavaScript, or at least server-rendered.

Check steps for the technical foundation

1. Load the home page with and without www, over http and https. Every variant must end at one single address.
2. View source on a category page, a product page and a filter page and see which canonical is set.
3. In Search Console, open the page indexing report: which groups of non-indexed URLs are large, and should they be?
4. Open your robots.txt and read every line. Cart, account, internal search and useless parameters belong there; images, CSS and JavaScript do not.
5. Compare the number of URLs in your sitemap with your number of products plus categories.
6. Measure one category and one product page on mobile and find out which element causes the largest paint.
7. Click through to your most important category from the home page and count the clicks.

5. Content and internal links

Which content is worth writing

Not every guide makes sense. It is worth it when three conditions line up: there is real search demand, the content helps with a buying decision, and you can link from there to relevant categories. Formats that work in stores:

• Buying advice and comparisons (what size do I need, A or B)
• Use, care, compatibility
• Answers to the questions customer service gets every day

Do not write guides on topics your category page would serve better. Otherwise both pages compete for the same query and Google decides which of the two weaker options to show.

Internal links as a ranking order

Internal linking tells Google which pages matter to you and distributes relevance. Three patterns that reliably work in stores:

1. Category to subcategory and back, with descriptive link text instead of here.
2. Guide to category, not to individual products, because products get discontinued.
3. Product to related products and to the parent category, so crawlers can get into the depth.

Make sure no important page is reachable only through internal search or the sitemap.

Snippets

Title and meta description decide click-through rate, not position. A pattern that works for stores: main query first, then one argument, then the brand. Avoid identical titles across hundreds of pages; with generated titles, build in at least one distinguishing field. When a page has plenty of impressions and almost no clicks, the snippet is the first suspect, and the full diagnosis is in impressions but no clicks.

6. Measuring: what is getting better and what is not

Without measurement, SEO is a belief system. Four figures are enough to start, all from Search Console and GA4:

A monthly comparison of the last 28 days against the same period a year earlier, split by category, product and guide pages, shows you which page type carries the store. That works with Search Console exports in a spreadsheet just as well as in a dashboard. How organic revenue fits next to your other figures is covered in the overview of the ecommerce metrics that matter.

Two additional metrics are worth tracking, because they show progress before revenue reacts:

• Coverage: for how many of your prioritized queries does a suitable, indexed page even exist? That ratio comes out of your keyword list and rises through work, not luck.
• Number of queries with impressions per page: if a reworked category appears for more distinct queries, Google understands it better, even when the position has not moved yet.

Also document what you changed and when. Without those notes you cannot tell six months later whether growth came from your work, from seasonality or from a change at Google.

Set realistic expectations. Technical corrections and better snippets often land within weeks. Structural and content work takes months, and nobody can promise a specific position. What you can control is coverage, the quality of those pages and the absence of technical errors.

Where to start

If you only have twenty hours, in this order:

1. Check indexing. Are cart, search and parameters out, and all categories and products in?
2. Fix redirects and errors. 404s that still get clicks, chains, mixed domain variants.
3. Rework your ten highest-revenue category pages. H1, intro, filters, internal links.
4. Rewrite titles and descriptions for those pages.
5. Create the missing subcategories where demand is demonstrable.
6. Give your bestseller product pages original copy and complete specifications.
7. Point internal links from guides and products at the important categories.
8. Improve load time at the biggest brakes, usually images and third-party scripts.

If you run on Shopify, a handful of platform specifics change the order slightly, and those are collected in the Shopify SEO guide. And if you also sell from a physical shop, none of the above covers the map pack: that runs on its own signals and has its own guide, local SEO for businesses with a storefront.

After that the boring, effective part starts: look at the same data every month, improve the next ten pages, close the errors. Ecommerce SEO rewards repetition more than ideas.