E-commerce SEO agency: steering SEO across a catalogue

A catalogue produces more URLs than there are search intents. That is where most of the SEO decisions on a merchant site come from: which pages exist, which ones are indexed, which ones receive authority, and in what order you touch them. This page is for the marketing director or the head of a merchant site who has to scope, then judge, an e-commerce SEO agency on a live catalogue.

It helps you decide which rules to set (facets, pagination, canonicals, page hierarchy), in what order to commit the work and what the delivered work is validated against. Here are the rules that hold when stock moves.

What a catalogue imposes on SEO

Organic search is still the leading channel for a merchant site: it accounts for 43% of e-commerce traffic (SEO.com, 2026), according to our SEO statistics, kept up to date. But that traffic plays out on particular ground, and three characteristics are enough to describe it.

• URL volume. URL generators (sorting, parameters, pagination, facets) continuously produce pages nobody decided to create. They absorb crawling and fragment signals across several near-identical versions.
• Page instability. Product pages can be “short-lived” (out of stock, end of line), whereas categories remain more durable anchors. A target that disappears or redirects too often makes part of the work void.
• Transactional intent. A ranking is only worth something if the page converts. Catalogue SEO is judged on revenue, not on an average position.

Hence three key success factors that come back in every decision below: indexing, granularity and stability. They govern the order of the work as much as its content.

The context adds pressure: a share of searches now ends with no click. On a catalogue, that shifts the effort towards the pages able to capture a buying intent and to be taken up as a reference, rather than towards piling up pages that exist without being asked for. The criteria for choosing an SEO agency in any context are settled beforehand; what follows is what a catalogue adds on top.

Catalogue architecture: categories, subcategories and hubs

In a catalogue, not all pages are equal. Category pages are more stable, more structuring, and can concentrate demand: they are the ones that carry SEO, and they then serve as hubs towards the products. Architecture is decided before anything else, because it conditions what the other workstreams can produce.

Depth, URL stability and platform constraints

Three constraints govern the architecture of a merchant site, and each has a direct consequence.

• Depth: pages that sit too deep receive authority poorly, even when you win external links
• URL stability: a target that changes (or redirects) destroys part of the value
• Selective indexing: indexing every product page is not always relevant (variants with no demand, duplicate content, long stock-outs)

An SEO agency works on existing shops, and the recurring points of attention concern templates and generated URLs:

• Consistency of URL paths: to avoid variants reachable through several routes
• Handling of parameters: sort and filter
• Stability of the category pages: they serve as hubs
• Control of redirects: during catalogue updates

These constraints can be described; they do not rank the platforms against each other: the same flaw is fixed differently depending on the stack, but it is fixed.

The four-level order of treatment

Setting this order once avoids re-opening every trade-off. It applies to content, to internal linking and to external authority.

Internal linking applies that order on a “category → subcategory → products” pattern, with contextual links from the guides and the comparisons, to strengthen the strategic category pages, to prevent several pages from competing on the same intent (cannibalization) and to reduce the depth of high-value products. The aim is to make crawling more intentional, not more voluminous.

Filters, facets and pagination: the indexing rule

This is the rule a catalogue calls for first, and the one that is least often written down. It is triggered as soon as volume, seasonality or the number of variants exceeds what the team can check by hand. Its deliverable is short: a list of URL families, a state assigned to each one, and the person who decides on new cases.

The states of a catalogue page

Filters and sort parameters quickly create a mass of URLs and an indexing debt:

• Risk of duplication
• Wasted crawl budget
• Fragmented signals: across several near-identical versions

Pagination, for its part, is not in the same family: it does not reorder content already served elsewhere, it gives access to the products that do not fit on the first page. Deindexing it on principle, or pointing it by canonical to page 1, amounts to cutting the path by which crawling reaches the depths of the catalogue. The discipline consists in deciding:

The delivery check that goes with this rule comes down to four verifications, to be passed before a page is considered handled.

• The page is indexable and actually indexed.
• The canonicals point to the right version, with each pagination page referencing itself and not page 1.
• The content and the internal linking reinforce the intended intent.
• The instruction used does produce the intended effect.

It is this last point that is missed most often, because three distinct mechanisms are treated there as interchangeable:

On a URL blocked by robots.txt, the directive placed in the page will never be read. Without that verification, a page has no technical capacity to capitalize on what it receives.

Which facets deserve an indexable page

A facet becomes a page in its own right when it meets three conditions:

• Real demand: expressed by queries you can name
• Stock that holds: enough available references for the page not to sit empty three weeks a quarter
• Enough depth: to justify its existence alongside its parent category

A facet that fails on any of the three stays available to the buying journey without being indexed.

Facet combinations are handled the same way, with a stricter threshold: every crossing multiplies the URLs, rarely the demand. By default, a single filter dimension is indexable; crossings only become so on an explicit, documented decision, with the query that justifies it. This rule is re-read at every change to the catalogue, not only when the workstream opens.

Product pages: content, evidence and structured data

Not all product pages deserve the same effort, and the question is not “should product pages be optimized” but “which ones justify a page of their own, and on what signal you switch”.

What deserves a page of its own, and what lives under its category

A product page justifies content of its own when it meets four criteria:

• A clear intent: product, benefits, uses
• Differentiating information: not a copy of the manufacturer’s
• A link to the relevant categories: the page is connected to them
• Proper conversion: otherwise you are funding traffic that does not buy

The others live under their category: present, reachable, correctly linked, but with no editorial investment of their own.

The switching signal is measurable: a product page that accumulates impressions on queries naming the product, or that carries a significant share of its category’s revenue, moves to individual treatment. Conversely, a variant with no demand or a reference in a long stock-out moves back down.

That movement is re-read every month, at the same time as the indexing rule. Setting this threshold once, in impressions or in share of category revenue, avoids deciding page by page: the list is recalculated, it is not re-argued.

Evidence, unique content and structured data

Three levers make a product page credible for a buyer as much as for an engine:

• Unique content: use cases, advice, differentiation
• Evidence: reviews, guarantees, a readable returns policy
• Structured data: product, price, availability, reviews

Adding customer reviews to a product page doubles the conversion rate (SiteW, 2026). Structured data, for its part, makes a page reusable and verifiable by systems that summarize, compare and recommend.

The same expectations hold upstream: source content able to be cited, useful and stable category pages, structured and credible product pages. One point of responsibility, finally: product visibility surfaces belong to a system separate from organic search, and on marketplaces optimization happens mostly within the platform’s constraints (content, titles, attributes, evidence, reviews). What you gain in authority on your own domain does not transfer mechanically to your marketplace pages.

Where authority should go, and in what order

External authority is one axis among others, and it is the one that costs the most when it is committed at the wrong moment or on the wrong page. Two decisions are enough to frame it: when you push, and on what.

Fix first, push afterwards

The execution sequence is the most structuring decision on a catalogue: fix indexing and structure first, push afterwards. On a merchant site this is particularly critical because of the URL generators, which can absorb the value if you strengthen pages that should not be indexed.

The preliminary diagnosis comes down to three questions:

• Which links already push categories, and which ones?
• Do the anchors reinforce understanding (product, category, brand) or do they create semantic noise?
• Do the best business pages receive authority, or is it captured by non-strategic pages (old products, useless facets, system pages)?

The third is the most profitable: it often reveals that there is nothing to acquire before redirecting what already exists. Link scarcity is the rule, so concentration on a few pages is the only reasonable lever.

The targets you do not push

A campaign loses much of its effectiveness if the targets change too often. In practice, you avoid pushing:

• Product pages in a long stock-out: with no alternative
• Very short collection pages: a few weeks
• Unstable URLs: parameters, multiple paths

Short-lived products are linked to the permanent categories. If they are removed, the rule is not an automatic redirect to the category:

Redirecting en masse to the parent category is a stopgap: the promise of the result is not kept, the signal sent to the engine is blurred, and too many regular redirects harm the crawling of the site.

The four-level order gives the sequence: pillar categories first, then subcategories and clean facets, then editorial hubs to capture upstream intent, then stable product pages case by case. On-site reinforcement of the target page (content, structure, trust elements, structured data) happens before or during the push, never after. The rule can be put in one sentence: you only strengthen what will carry the same URL in a year.

What an agency delivers, and who approves what

A serious engagement produces actionable deliverables. On a catalogue, expect at minimum:

• An authority audit (inbound links, anchors, target pages, risks)
• A prioritized roadmap (targets, pace, technical prerequisites, validation criteria)
• Monthly reporting geared to impact (visibility, rankings, conversions), not just “number of links”
• Governance (who approves the targets, who updates the pages, who fixes indexing drift)

On the level of evidence expected and the prioritization logic, what an SEO and GEO audit has to deliver gives the frame. A good partner also documents:

• The hypotheses: why these pages, why this pace
• The risks: anchors, unstable pages, duplication
• The evidence and the progress: Search Console, Analytics
• The corrective actions: if the impact does not come

Those four requirements are the criteria an agency on a catalogue is judged on, month after month.

Who approves what, and what stays with you

The scope is set at kick-off, in five lines:

• The list of eligible pages: and those excluded
• The indexing rules: facets, pagination
• The monthly pace
• The validation criteria: Search Console and Analytics
• Who does what: content, dev, approval, publication

That last line deserves detail, because it is what decides the real workload. Producing category content can be delegated, but approving commercial arguments and stock information stays with the advertiser. Deploying a canonical rule or an indexing tag goes through the team that owns the template: the agency specifies, someone puts it into production, and sign-off is contradictory.

Deciding on a stock-out (redirect, unpublish, keep with an alternative) belongs to whoever knows the restocking, which is rarely the provider. On large catalogues, the difficulty is not “finding ideas”, but prioritizing: naming who decides those three cases is worth more than an org chart.

Measuring: visibility, business, generative presence

Three families of indicators are enough, and they are read together.

• Visibility: impressions (surface gained), clicks and CTR (snippet quality and fit with the intent), rankings, in particular progress towards the top 3
• Business: conversion rate by page and by category, organic revenue and, ideally, margin, revenue per visitor and journey (mobile versus desktop)
• Generative presence: the presence and reuse of your pages as sources in assistants’ answers, which favours “source” pages (guides, enriched categories, structured product pages) able to be taken up, not only to rank

A fourth monitor is almost always missing: the catalogue non-regression check, that is, the ratio of useful indexed pages to useless indexed ones. An indexing rule is not validated on an intent, it is validated on that ratio.

Two drifts are watched every month: pages indexed with no impression at all over three months, and business pages that have left the index with no decision. The first signals a rule that is too permissive, the second a rule that is too aggressive or a deployment accident. Without that check, a rise in rankings can mask an indexing debt that will cost more than the gain.

If the question before the steering committee is tying organic revenue to spend and holding a horizon, calculating and steering SEO ROI gives the method.

Depending on your situation, three extensions follow:

• You are opening a project rather than steering a live site: in a redesign or a build, the architecture foundations, the redirect plan and the acceptance check before go-live are decided upstream; see what an e-commerce agency secures on the SEO side during a redesign.
• You are entrusting the engagement to one person rather than a team: what the engagement of an e-commerce SEO consultant covers, and what it leaves with you, is scoped before signing.
• You need to strengthen the authority of your categories: the sources, the anchors, the pace and the risk controls are the business of what an off-site SEO agency brings to acquiring external authority.

To establish the catalogue’s starting point (indexing by URL family, canonicals, pages that receive authority) before deciding on the targets, the 360 SEO and GEO audit provides that diagnosis.

FAQ: e-commerce SEO agency

What does an e-commerce SEO agency do that is specific?

It works on a catalogue, not on an editorial site: URL volume, facets and pagination, pages that appear and disappear with stock. Its main contribution is the order of the work: it arbitrates the targets (categories, subcategories, supporting content, products), sets the indexing and sequencing rules, and ties the investment to organic revenue rather than to an average position.

Which pages should be indexed first on a catalogue?

Categories, which are more stable and more structuring, then the subcategories and facets that combine real demand, lasting stock and sufficient depth. Product pages are indexed selectively: those that carry a named intent and convert. Sorts and parameters are consolidated or blocked; pagination stays crawlable and self-canonical, because it gives access to the depths of the catalogue.

How long does it take to see results?

The delay varies with the competition, the quality of the target pages and the site’s ability to be properly crawled and indexed. On a catalogue, indexing fixes produce the first signals before ranking gains: which is precisely why they come first in the sequence.

How do you judge an e-commerce SEO agency’s reporting?

On its ability to tie actions to measurable variations. In Search Console: impressions, clicks, CTR, position, the pages and queries that are progressing, and the indexing state by URL family. In Analytics: organic conversion rate, revenue, average basket if available, performance by device and by category. Counting links or rankings without that connection does not allow a judgement.

What is the link between e-commerce SEO and GEO?

Generative systems compare and recommend from clean, verifiable information. The pages that count are therefore the “source” pages: guides, enriched categories, product pages with unique content, evidence (reviews, guarantees) and structured data. The same work that makes a product page credible for a buyer makes it citable by an assistant; unstable or unreliable pages are rarely taken up.

What stays on the client side when you delegate your catalogue’s SEO?

Approving commercial arguments and stock information, putting the indexing rules into production through the team that owns the template, and deciding on stock-outs (redirect, unpublish, keep with an alternative), which depends on restocking. Those three decisions cannot be delegated; specifying and producing category content can.