2,500 sneakers, one product template
One of South Africa's most established sneaker retailers, a catalogue of around 2,500 products, and a hard requirement that the team could run the store themselves afterwards. We built it on a single dynamic product template inside a fixed 100-hour boundary, and hit the date.
- ~2,500 products, rebuilt on Shopify Online Store 2.0 without a template per product — everything variable lives in metafields.
- The Lemkus team used Claude to research and structure product data into those metafields, which is what made a catalogue that size tractable.
- Fixed 100-hour boundary, hit on schedule, paid in full before launch. The mobile nav was rebuilt late on client feedback, not on a metric.
The brief
Lemkus is one of South Africa's most established sneaker retailers, with a catalogue of roughly 2,500 products.
The requirement that shaped everything else wasn't a design one. It was that after launch, the Lemkus team had to be able to run the store themselves — without calling a developer to change a layout, and without inheriting a storefront only the person who built it could edit.
The template problem
At 2,500 products, the obvious approach — a template per product type, then another for the exceptions — produces a store nobody can maintain. It solves today's layout and creates a permanent tax on every future change.
So there is one product template. Everything that varies between products — specifications, colourways, copy, supporting content — lives in Shopify metafields and renders dynamically into shared sections. Layout structures stay inside theme-native settings and sections wherever possible, so the team can change the store from the admin rather than from a code editor.
That decision creates a different problem, which is that a metafield structure is only as good as the data in it. Populating structured data across thousands of SKUs is the kind of work that quietly kills a migration.
The catalogue build
Rather than manually researching and writing product information across thousands of SKUs, the Lemkus team used Claude to help find, structure and prepare that information for the relevant Shopify metafields.
Worth being precise about who did what: we built the structure, they filled it. That split is the point. A metafield architecture the client can populate themselves is a different asset to one that depends on us for every new season.
What changed after sign-off
The initial mobile navigation worked. There was nothing technically wrong with it. But once the developed site was in front of the Lemkus team and being used properly, it didn't match the quality of the rest of the experience. So we revisited it.
That happened because the client was able to experience a working product rather than predict every interaction from a static design. It's an argument for building in a way that keeps things editable after sign-off, rather than treating a signed design as final.
What shipped
The boundary
The project ran against a fixed 100-hour boundary and a frozen design. Both existed so the scope conversation happened once, at the start, rather than continuously.
It landed on schedule and was paid in full before launch. Some things were deliberately pushed out rather than squeezed in — a live shopping and launches section was scoped as a V2 rather than allowed to expand the build.
Knowing what to leave until later is most of what a fixed boundary buys you.
Put your store through the same read.
Every engagement starts with a free store diagnostic. Send the URL and we'll come back with the specific things costing you orders, ranked by what they're worth — whether or not you ever work with us.
Related case studies
All case studiesSA Apparel Retailer
Apparel · South Africa
The wholesale operation was already working. The opportunity was to make it easier to manage.
Root4
Skincare · South Africa
The goal was never to make it look like a Shopify store. It was to use Shopify as the engine while preserving what made the brand distinctive.
SA Apparel Retailer
Apparel · South Africa
A campaign doesn't fail on the big idea. It fails on the payment method that quietly stopped working eleven days out.