Project Description
Technical and structural optimization of a four-brand Shopify network
Four shops, four sports, a single commercial machine behind it. Baseball Town, Volleyball Town, Pickleball Town, and Dek Hockey Town share the same Shopify infrastructure under the Sports Town Group banner, with a cross-catalog, common brands, and a multi-buy funnel. This configuration has many advantages on the operations side. On the SEO side, it creates a series of problems that not many Shopify installations encounter.
The mandate focused precisely on these problems: to ensure that four sister sites, built on a shared base, are read by the engines as four distinct entities, each legitimate on its sport, without stepping on each other or diluting the signals of the whole, with the focus being a GEO performance , therefore, for the response engines, otherwise known as AIs and LLMs.
A multi-store setup that doesn’t treat like a regular site
The first task was to map what is really shared between the brands and what must remain specific to each one. Templates, sections, content blocks, breadcrumbs, headers, footers, collection structure: on an architecture of this type, a change made in one theme propagates to several storefronts at once. It’s powerful, and it’s also the fastest way to duplicate an error by four.
The approach adopted was therefore to reason in layers. What is part of the common technical base has been standardized once and for all. What carries the search identity of each brand, i.e. the signals that tell Google that we are talking about baseball bats and not pickleball rackets, has been isolated and processed shop by shop. This separation conditions everything else.
JSON-LD: Rebuild consistent structured data across the network
Shopify generates basic markup. It is rarely sufficient on a catalog of this size, and it becomes frankly insufficient when the same product can appear in several brand contexts.
The JSON-LD markup has been thoroughly reused: product sheets, collections, organizational units, prices and availability, notices when they exist. The objective was not to tick boxes in Google’s test tool, but to build a readable graph, where each product is unambiguously linked to the right brand, the right sport and the right commercial offer. A Rawlings bat sold at Baseball Town and a bat listed elsewhere in the group don’t tell the same story to engines, and the markings should reflect that distinction.
The bilingual constraint is added on top of that. Each brand exists in French and English, which doubles the number of entities to be declared cleanly and imposes rigour on language correspondence.
Catalog-wide metadata
On a sports e-commerce, titles and descriptions are rarely written by hand one by one. The question is not to write beautiful guidelines, but to conceive the logic that generates them.
The work consisted of establishing metadata patterns differentiated by type of page and by brand, taking into account how players actually search. A parent looking for a glove for their child, a softball player comparing two models of sticks, a dek hockey fan looking for a specific brand: these intentions have neither the same vocabulary nor the same granularity. The patterns were calibrated accordingly, and then the high-stakes pages were taken over individually.
This type of trade-off between systematization and manual processing is found in most of our content optimization mandates applied to large catalogs.
Snippets and presentation in results
Once the structured data was clean, all that remained was to work on what the engines actually displayed. Price, inventory, variants, price ranges, enriched elements: each signal sent directly influences the click-through rate, and in a market where the same product is sold by several Canadian retailers, this detail weighs heavily.
The work focused on the reliability of the information reported, its freshness, and its consistency between the displayed version and the declared version. Nothing damages trust faster than an outdated price in a profit and loss statement.
Site structure: prioritizing without cannibalizing
The most delicate point of the mandate. Four brands that share brands, types of products and sometimes pages is fertile ground for internal cannibalization.
The architecture of the collections has been revised to establish a clear hierarchy: sport, category, subcategory, brand, then product. Each level has a reason for existence and a corresponding research intent. Redundant collections were merged or removed, those with real potential were strengthened, and canonicals were clarified wherever two paths led to the same content.
This prioritization is based on a detailed reading of the search behavior in each discipline, which is directly related to the SEO strategy before being a technical issue. The same reasoning guided the recovery of Rose Boréal’s Shopify collections, in a different context but with the same basic mechanism: collection pages are the real landing pages of an e-commerce, and they are treated as such.
One legible network, four distinct identities
At the end of the mandate, the four brands are based on a unified technical base while sending differentiated signals. The markup is consistent, the metadata follows a logic that is sustainable over time, and the structure of the collections allows each store to position itself on its own ground without borrowing that of the other three.
A multistore setup is never SEO-neutral. Well managed, it makes it possible to pool technical efforts while multiplying semantic coverage. Poorly managed, it produces four sites that compete with each other. The whole difference lies in the rigour with which we separate what should be common from what should remain unique.
