The Complete Guide to eCommerce SEO
eCommerce SEO is not blogging with products attached. It is the work of matching a catalogue to the way people actually search, deciding which of the thousands of URLs your platform can generate deserve to exist, and then proving the result in contribution rather than in rankings.
CHAPTERS
8 chapters
READING TIME
38 min read
LAST UPDATED
18 September 2026
LEVEL
Intermediate
The page that decides which product gets seen is usually the least-optimised page on the store.
EXECUTIVE SUMMARY
What this guide argues, in five points
Almost every eCommerce SEO problem we are asked to fix turns out to be one of five things, and none of them is a missing plugin. They are structural: the wrong pages exist, the right pages are empty, the platform is generating URLs nobody decided on, the catalogue is described in internal language, and the reporting cannot tell growth from brand demand.
- Category pages, not product pages and not the blog, carry the commercial search demand. On most stores they are the thinnest pages on the site.
- Keyword research for a catalogue is demand mapping, not a list. The output is a set of pages that should exist, in priority order.
- Faceted navigation is the largest untapped opportunity and the largest crawl-waste risk on the same site. Decide per facet, never globally.
- Product content has to be prioritised by revenue rank. Two hundred pages done properly beats four thousand done badly.
- Report non-branded landings by template against contribution. Total organic sessions hides everything that matters, including whether you are growing at all.
The eight chapters below are the detail behind those five points, in the order we work through them when we take over organic search on a store.
- Contents
Eight chapters
Read it in order the first time. After that, the chapter index is the useful part.
01
A store is a small number of templates producing thousands of URLs. That single fact creates every problem in this guide.
4 min
02
The output of keyword work on a store is a set of pages that should exist, ranked by the revenue behind them.
5 min
03
Commercial demand lands on category pages. On most stores those are the thinnest pages on the site.
6 min
04
Filters are the biggest opportunity and the biggest crawl-waste risk on the same site. Decide per facet, never globally.
5 min
05
You will not write four thousand descriptions. Rank by revenue, write at the top, structure everything else.
5 min
06
Five things matter at catalogue scale. The other 195 items in the audit are findings, not work.
5 min
07
Supporting content should feed the category layer, not compete with it. Increasingly it also decides whether assistants quote you.
4 min
08
Brand demand lands in organic. Until you separate it, the report tells you nothing about whether the work is working.
4 min
- Before you start
Who this guide is for
Written for people who own organic performance on a store, whether or not they do the work themselves.
Founders and heads of eCommerce
You want to know which parts of eCommerce SEO are real, which are sold to you, and what a reasonable timeline looks like.
In-house SEO and merchandising teams
You control the catalogue and the category tree and want a defensible order of work rather than a 200-item audit.
Developers inheriting a store
You need to know which platform defaults are quietly generating URLs nobody decided on before you change anything.
NOT WRITTEN FOR
People looking for a plugin that does SEO. There is not one. The work in this guide is demand mapping, merchandising and catalogue data — most of it done by a person who understands the products.
CHAPTERS
- 01 Why eCommerce SEO is a different discipline
- 02 Demand mapping, not keyword lists
- 03 Category architecture: the layer that earns
- 04 Faceted navigation and the indexation decision
- 05 Product pages at catalogue scale
- 06 Technical foundations that actually matter
- 07 Content, links and AI citation
- 08 Measuring eCommerce SEO honestly
CHAPTER 01
4 min
Why eCommerce SEO is a different discipline
A store is a small number of templates producing thousands of URLs. That single fact creates every problem in this guide.
Most SEO advice is written for sites with a few dozen pages that each say one thing. A store is the opposite: a small number of templates generating thousands of URLs, where the same product legitimately belongs in four places and the platform will happily invent a fifth. The problems that follow are structural, and they do not appear on a content site at all.
Problem
Why it is specific to eCommerce
Where it is handled
Thousands of near-duplicate URLs
Filters, sorts, variants and pagination multiply one page into hundreds
Chapter four
The best pages are templates, not documents
A category page is generated, so it is only as good as the data behind it
Chapter three
Content at catalogue scale
Nobody is writing four thousand original product descriptions
Chapter five
Inventory changes the page
A page that ranked can go out of stock and quietly become worthless
Chapter six
Brand demand masks everything
Total organic looks healthy while non-branded acquisition falls
Chapter eight
If your SEO plan does not name these five, it is a content plan with a search label on it.
KEY TAKEAWAY
The blog is not where eCommerce SEO is won. It is a supporting asset. The commercial demand lands on category pages, and that is where the work belongs.
The two questions worth asking first
- Does the page that should rank actually exist? On most stores a third of the addressable demand has no page pointed at it at all.
- If it exists, is there anything on it? A grid of products with a heading is not a page; it is a query result with a title tag.
Those two questions produce more revenue in the first quarter of an engagement than any technical audit we have ever run. The technical work matters, but it matters second, because a perfectly crawlable empty page is still an empty page.
CHAPTER 02
5 min
Demand mapping, not keyword lists
The output of keyword work on a store is a set of pages that should exist, ranked by the revenue behind them.
Keyword research for a catalogue is not a list of terms with volumes next to them. It is a map of how buyers describe what you sell, converted into a set of pages that should exist, ranked by the revenue behind them. The deliverable is a site structure, not a spreadsheet tab.
Start from the catalogue, not the tool
1
Export the catalogue with its attributes. Type, material, size, use case, brand, compatibility. These attributes are the vocabulary buyers search in.
2
Collect the phrasing that already works. Internal site search, existing non-branded landings, marketplace search suggestions and the questions support answers repeatedly.
3
Cluster by intent, not by string similarity. Two phrasings that return the same results page are the same page, whatever the tool says.
4
Read the results page for every cluster you intend to build. If it returns articles, a category page will not rank there and you should not build one.
5
Attach a revenue estimate. Volume multiplied by a realistic click share, conversion rate and contribution per order. This is what ranks the work.
KEY TAKEAWAY
Search intent is decided by the results page, not by the words. The fastest way to waste a quarter is to build a commercial page for a query that returns nine explainers.
The three page types demand maps to
Demand shape
Page type it needs
Example phrasing
Category
A category or collection page
A product type, sometimes with one attribute
Attribute or use case
A sub-category or curated collection
Product type plus material, size, compatibility or occasion
Question or comparison
A guide, comparison or buying page
How to choose, which is better, what size do I need
The third row supports the first two. It should link into them deliberately rather than sit in a blog archive on its own.
The most common output of this chapter is uncomfortable: a list of twenty to forty pages the store should have and does not, most of them one level below the categories in the main navigation. That list is the SEO plan for the next two quarters.
CHAPTER 03
6 min
Category architecture: the layer that earns
Commercial demand lands on category pages. On most stores those are the thinnest pages on the site.
The category layer is where commercial search demand lands and where most stores have done the least work. A buyer searching a product type is not looking for one product; they are looking for a shortlist, and the page that presents that shortlist is the page the search engine is choosing between.
The category page is a merchandising decision and a content decision at the same time. Most stores make only the first.
What a category page actually needs
- A heading in the customer’s language. Not the internal merchandising name, not a season code. The phrase buyers use, which chapter two just gave you.
- A short orienting paragraph above the grid. Two or three sentences that tell somebody who has just arrived what is here and how to choose.
- Buying guidance below the grid that answers a real decision — sizing, compatibility, materials, what to avoid. This is the part almost nobody writes.
- Links out to the narrower pages that deserve to exist, in the body copy rather than only in a sidebar filter.
- In-stock products in the first row. The single most common silent conversion failure on an otherwise good category page.
- A written title tag and description. Generated from the category name is a default, not a decision.
KEY TAKEAWAY
If you did only one thing from this guide, rebuild the top twenty category pages by revenue against that list. It is the largest single source of upside on most stores we take over.
Depth: how far down to build
The temptation is to create a page for every attribute combination the catalogue allows. The discipline is to create one only where three things are true at once: there is demand, there is enough stock to make the page worth landing on, and there is something genuinely different to say. Two out of three is not enough, and a page that fails all three is crawl waste with a nice URL.
Level
What it is
Build it when
Department
The top of the tree, usually in the main navigation
Always. It is the hub the rest links from.
Category
A product type buyers search by name
Always, and this is where the effort goes
Sub-category
Product type plus one attribute
Demand, stock depth and something to say — all three
Two attributes
Product type plus two attributes
Rarely. Only where demand is genuinely proven.
Three or more
Long attribute chains
Effectively never. This is filter territory.
Most stores are missing the third row entirely and have accidentally published thousands of the fifth.
Internal linking in three directions
Internal linking is the part of eCommerce SEO that costs nothing and is almost always absent. Three directions matter, and only the first is usually present.
- Down. Department links to category links to sub-category. Usually handled by the navigation, and usually the only direction anybody builds.
- Sideways. Related categories linked to each other from body copy, where the relationship is real rather than merchandised.
- In. Guides, comparisons and buying pages linking into the category they support, with the phrasing buyers actually use as the anchor text.
EXPERT VIEW
“When a store tells us its blog is not converting, the problem is almost always that the blog is ranking for terms the category pages should own, and the category pages are ranking for nothing at all.”
Dazzle Commerce · Organic search team
Our article on why category pages are undervalued is the short version of this chapter, with the reasoning behind it.
CHAPTER 04
5 min
Faceted navigation and the indexation decision
Filters are the biggest opportunity and the biggest crawl-waste risk on the same site. Decide per facet, never globally.
Filtering is the feature that makes a large catalogue usable and the feature that generates more indexable URLs than the rest of the site combined. Indexing all of them and indexing none of them are both wrong, and both are common. The decision is per facet, and it is a decision somebody has to actually make.
Three tests, applied per facet
1
Is there demand? Do people search this combination, phrased this way, with commercial intent.
2
Is there stock depth? A filtered page returning four products is a bad landing page whatever it ranks for.
3
Is the page stable? A combination that empties every fortnight will lose whatever it earns.
Facet combination
Index it?
Why
Category plus attribute
Yes, where demand exists
Matches how people search and stays stable over time
Category plus brand
Usually
High commercial intent and a genuinely different product set
Category plus two attributes
Selectively
Only where volume and stock depth both justify it
Category plus price band
No
Volatile, thin, and rarely how a search is phrased
Category plus sort order
Never
The same products in a different order — a duplicate by definition
Internal search results
Never
Infinite, low quality, and a classic crawl trap
Anything you do not index still needs a canonical and a crawl directive. Undecided facets are where crawl budget quietly disappears.
KEY TAKEAWAY
A faceted page you decide to index should be treated as a real page: a written heading, copy, a title tag and internal links. If it does not deserve that effort, it does not deserve to be indexed.
Pagination, variants and the other duplicate sources
- Pagination. Let the pages be crawled and self-canonical. Canonicalising page two to page one hides the products on it.
- Variants. One page per buying decision. If colour is a separate purchase decision in your category, it may earn a page; if it is a swatch, it should not.
- Session, tracking and sort parameters. These should never have been indexable. Check what your platform and your apps are appending.
- Out-of-stock and discontinued products. Keep the URL, keep the content, show the alternatives. Deleting it throws away everything the page earned.
Platforms differ in how much of this you control. On a hosted platform the prefixes and parameter behaviour are often fixed, which changes the tactics but not the decision. Our Shopify SEO guide covers the platform-specific version of this chapter in detail.
CHAPTER 05
5 min
Product pages at catalogue scale
You will not write four thousand descriptions. Rank by revenue, write at the top, structure everything else.
Nobody is going to write four thousand original product descriptions, and pretending otherwise is why product content projects stall in month two. The workable approach is to rank the catalogue by revenue, write properly at the top, structure everything else, and let the structure do the work at the tail.
Three tiers, three levels of effort
Tier
Share of catalogue
Treatment
Revenue top decile
Roughly 10%
Written by a person: original copy, full specification, own imagery, FAQs from real questions
The middle
Roughly 40%
Structured: complete attributes, consistent specification tables, unique opening paragraph
The tail
The rest
Data-complete and indexable, with attributes filled in and no duplicate manufacturer blurb as the only content
The order matters more than the ambition. Two hundred pages done properly beats four thousand done badly, and it finishes.
KEY TAKEAWAY
Product attribute data is content. A complete, accurate specification table earns more search visibility on a technical product than three paragraphs of adjectives.
What a product page has to carry
- A unique opening paragraph that says what it is and who it is for, not a manufacturer blurb duplicated across every retailer selling it.
- Complete, structured attributes. Dimensions, materials, compatibility, care, what is in the box. This is what comparison and AI answers read.
- Delivery and returns stated on the page, not two clicks away in a policy document. It is the most common unanswered question at the point of decision.
- Real questions answered. Pull them from support tickets and reviews rather than inventing them.
- Valid, singular structured data. One Product block with price and availability. Duplicate markup from theme and apps is the most common markup bug we find.
Our short guide on the nine product page elements that move conversion is the on-page counterpart to this chapter.
CHAPTER 06
5 min
Technical foundations that actually matter
Five things matter at catalogue scale. The other 195 items in the audit are findings, not work.
Technical SEO on a store is not a 200-item audit. Five things reliably matter at catalogue scale, and the rest of the audit is usually a list of findings nobody will ever action. Fix the five, then stop.
Measure the templates that carry revenue, on the devices customers actually use.
The five that matter
1
Crawl budget going somewhere useful. If a large share of crawling lands on filters, search pages and parameters, the pages you care about are being visited less often than they should be.
2
Indexation matching intent. Every URL type on the site should have a deliberate answer to whether it may be indexed. Most stores have never written that list down.
3
Speed on the revenue templates. Category and product, measured on a mid-range phone on a real network. The home page in a lab tells you nothing useful.
4
Rendering. If the product grid, the price or the copy only exists after JavaScript runs, verify what is actually in the rendered HTML rather than assuming.
5
Structured data that is complete and singular. Product, breadcrumb and organisation, validated on the live templates rather than in a staging environment.
KEY TAKEAWAY
Technical problems on a store are almost always the theme, an app or the platform’s defaults. They are rarely the thing your SEO tool ranked as critical.
Two that get forgotten
- Out-of-stock handling. A product that ranks and goes out of stock is a page that quietly stops converting. Decide what the page shows and whether it still links onward.
- Migrations and redirects. More organic revenue is lost in replatforming than in any algorithm update. Map every URL before the change, and check every redirect resolves in one hop afterwards.
Review these quarterly rather than continuously. Technical SEO on a store is not a monitoring exercise; it is a short list checked at a sensible cadence, with the rest of the time spent on the category layer.
CHAPTER 07
4 min
Content, links and AI citation
Supporting content should feed the category layer, not compete with it. Increasingly it also decides whether assistants quote you.
Content and links are the supporting half of eCommerce SEO. They are not where the commercial demand lands, but they are what makes a category page credible enough to hold a position against retailers with more authority than you, and they are increasingly what decides whether your products are cited in an AI answer at all.
Content that supports a catalogue
- Buying guides that answer a real decision — sizing, compatibility, what to avoid — and link into the categories they support.
- Comparisons where two options genuinely compete. These earn links and answer a query a category page cannot.
- Specification and reference content for technical categories. Unglamorous, highly linkable, and the kind of page assistants quote.
- Nothing that competes with your own category pages. An article ranking for a commercial term instead of the category page is a loss dressed as a win.
KEY TAKEAWAY
Every supporting article should name the category page it exists to support. If it cannot, it is probably competing with one.
Links, honestly
Most stores will not out-link a marketplace or a national retailer, and planning as though you will is how budgets get wasted. What works is narrower: digital PR around data you actually have, supplier and stockist relationships, category expertise that specialist publications will cite, and product pages good enough to be linked to on their own merits.
Being cited by AI assistants
Assistants and AI overviews increasingly sit between the search and the store. What gets quoted is structured, specific and verifiable: complete specifications, clear pricing and availability, stated delivery and returns terms, and reviews that read like people rather than marketing. The work overlaps almost entirely with chapter five, which is the good news.
Our article on getting products cited by AI assistants covers what these surfaces actually read and what they ignore.
CHAPTER 08
4 min
Measuring eCommerce SEO honestly
Brand demand lands in organic. Until you separate it, the report tells you nothing about whether the work is working.
Organic search is the easiest channel to report dishonestly, because brand demand created by every other channel lands in it. A store can run a flat non-branded acquisition line for a year while the organic chart climbs, and nobody notices until paid media is cut and the organic number falls with it.
Report by template and by intent. Total organic sessions is the number that hides the problem.
The four views worth building
View
Question it answers
Cadence
Non-branded landings by template
Is the category layer actually compounding?
Monthly
Indexed pages versus intended pages
Is the site publishing URLs nobody decided on?
Monthly
Organic contribution, not revenue
Is the traffic landing on products worth having?
Monthly
Page-level movement, not keyword rank
Which pages are gaining, and why?
Quarterly
A category page winning fifteen mid-tail terms matters more than a rank on one head term, and only the first view shows it.
KEY TAKEAWAY
If branded and non-branded are not separated in your reporting, you cannot tell growth from harvesting, and the SEO report will look healthy for as long as brand demand holds up.
What to expect, and when
- Weeks one to six: technical and indexation changes land. Crawl patterns move. Revenue mostly does not.
- Months two to four: rebuilt category pages start ranking on mid-tail terms. This is the first honest signal.
- Months four to nine: the compounding phase, where internal linking and content depth begin to carry newer pages faster.
- Anything faster than that is usually brand demand, a seasonal effect, or a reporting artefact. Check before celebrating it.
Give architecture work a full quarter before judging it. Crawling, re-indexing and a complete demand cycle all have to happen before the number means anything, and cutting the work at week six is the most reliable way to pay for all of the cost and none of the return.
- Worked examples
Three decisions, before and after
The three arguments that come up on almost every store we take organic search on.
CATEGORY HEADING
Merchandising language versus demand
Before
H1: “AW26 Layering Edit”
Title tag: “AW26 Layering Edit | Brand”
After
H1: “Merino Base Layers”
Title tag: “Merino Base Layers | Brand”
The internal season code carries no demand. The page still merchandises the edit; it simply stops describing itself in a language only the buying team speaks.
CONTENT STRATEGY
The blog outranking the category page
Before
An article ranking sixth for the main commercial term; the category page nowhere
After
Article refocused on the buying decision and linked into the category, which is rebuilt to rank
Sending commercial traffic to an article adds a click before the shortlist. Both pages can exist; only one of them should be trying to win the commercial query.
INDEXATION
Indexing every filter, or none
Before
All filter combinations indexable, including price bands and sort order
After
Category plus attribute indexed where demand and stock exist; price, sort and multi-facet blocked
Indexing everything wastes crawl budget on thin duplicates. Indexing nothing throws away the most commercially valuable pages on the store. The decision is per facet.
Anonymised from stores we manage. The shapes recur; the specifics will not match your catalogue exactly.
- Take these with you
Three checklists
Written to be worked through rather than read. Nothing here needs a tool you do not already have.
Before you start
Half a day, once, then after any catalogue change.
- Category demand mapped to pages that should exist, ranked by revenue
- Branded and non-branded separated in reporting
- Every URL type on the site has a written indexation decision
- Top twenty category pages by revenue identified
- Catalogue ranked by revenue into three content tiers
- Results pages read for every cluster you intend to build
Every category page
Run this on the top twenty by revenue.
- Heading matches how customers phrase the category, not the merchandising tree
- Short orienting paragraph above the grid
- Buying guidance below the grid that answers a real decision
- Links out to the narrower pages that deserve to exist
- First row of the grid is in stock
- Title tag and meta description written, not generated
Technical, quarterly
Twenty minutes, four times a year.
- Crawl stats checked for budget going to filters and search pages
- Indexed page count reconciled against intended pages
- Category and product templates measured on a mid-range phone
- Rendered HTML checked for grid, price and copy
- Structured data validated on live category and product templates
- Out-of-stock and discontinued handling confirmed
Copy these into your own tracker. There is no download form and no email wall.
- Frameworks
Two frameworks worth keeping
The two decisions that get reargued every quarter, written down so they stop being reargued.
FRAMEWORK ONE
The page-should-exist test
Before building any new category, sub-category or filtered page, run it against four tests. A page that fails any one of them is crawl waste with a nice URL, however good the volume looked.
Test
Pass condition
If it fails
Demand
People search this, phrased this way, with commercial intent
Do not build it
Intent match
The results page returns retailers, not explainers
Build a guide instead
Stock depth
Enough products to make a useful shortlist
Keep it as a filter, not a page
Something to say
There is real guidance for this specific buyer
It will be thin; leave it out
Four passes and the page earns a written heading, copy, a title tag and internal links. Fewer than four and it earns a canonical.
The same test applies to faceted URLs. A filter you index is a page, and it should be held to the same standard.
FRAMEWORK TWO
The eCommerce SEO priority order
When everything is broken, the order matters more than the list. Each step assumes the one above it has cleared, because work at step five is wasted while step one is wrong.
1
Measurement integrity. Branded and non-branded separated, templates reportable. Without this you cannot tell whether anything worked.
2
Indexation control. A written decision for every URL type. This is where crawl budget is won and where most stores have never decided anything.
3
The category layer. Top twenty pages by revenue rebuilt. The largest single source of upside on most stores.
4
Internal linking. Down, sideways and in. Three directions that cost nothing and are usually absent.
5
Product content by revenue tier. Top decile written, the middle structured, the tail data-complete.
6
Technical performance. Theme, apps and rendering, measured on the templates that carry revenue.
7
Supporting content and links. Guides and comparisons that feed the category layer rather than competing with it.
International sits outside this order. If you are opening a market, settle the URL and hreflang model before step one.
- In summary
Key takeaways
Eight sentences you could hand to somebody who is not going to read the guide.
1
Category pages carry the commercial demand. Product pages convert it. Optimising only the second means paying for traffic you could have earned.
2
Keyword work on a catalogue produces a list of pages that should exist, ranked by revenue — not a spreadsheet of terms.
3
Search intent is decided by the results page, not by the words. Read it before building a commercial page.
4
Decide facet indexation per facet. Indexing everything and indexing nothing are both expensive mistakes.
5
Rank the catalogue by revenue and write at the top. Two hundred product pages done properly beats four thousand done badly.
6
Five technical things matter at catalogue scale. The other 195 audit findings are findings, not work.
7
Supporting content should feed the category layer. An article outranking your own category page is a loss dressed as a win.
8
Separate branded from non-branded or the organic report will look healthy for as long as brand demand holds up.
If you want a single first action: open your top three category pages and read them as a customer who has never heard of you.
- Go deeper
Where this work sits in the practice
Three services that pick up the chapters above.
Demand mapping, category architecture, technical SEO and content built around commercial intent.
Theme work, rendering, app rationalisation and the performance improvements chapter six depends on.
A technical and organic review of your store, with findings ranked by revenue impact rather than severity.
- Keep reading
Related guides
The rest of the guide library, all published in full.
Complete guide · 32 min
The platform-specific version: URL prefixes, tag filters, theme duplication and app-driven markup.
Complete guide · 37 min
Where organic sits in the wider channel mix, and how to tell whether it is the constraint at all.
Complete guide · 37 min
The category pages from chapter three, treated as a conversion problem as well as a ranking one.
Want this run on your store rather than read about it?
Our organic search engagements open by mapping category demand against the pages you actually have, then rebuilding the layer that carries it.