Boardside is a Latvian watersports retailer - wetsuits, surfboards, kites, wakeboards - with a shop in Riga and a warehouse behind it. Their store ran on a custom-built system written years earlier, with no update path and no one left who knew it. We moved the business onto Shopify: 1,914 products, 244 categories, ten years of accumulated search visibility, and two languages now on two domains.
The design came as a Shopify theme. Our work was everything else - getting a decade of catalogue out of a system with no real export, and making the new store behave like a Latvian shop rather than a translated American one.
A migration nobody had to clean up afterwards
Moving off a bespoke platform is not an export and an import. The old data model was whatever its author needed at the time: attributes as free text, spellings that drifted over years, two different answers to the returns period.
So the catalogue was rebuilt as structured data on the way across. Board type, fin set, binding type and size became defined values in both languages - which is what turns a week of work into an afternoon later on: when the client sent back a spreadsheet of Latvian corrections, 962 of them went live in one pass.
The search visibility came too. Over 2,900 redirects map every old address to the right page, in the right language, on the right domain - including Latvian product and category URLs that differ from their English equivalents.
Latvian at the root, English on .eu
Latvian sits at the root of boardside.lv and English on boardside.eu, each domain serving one language cleanly. Shopify’s own language switcher cannot cross between two domains, and the usual fix is to rent one from the app store every month.
Instead the switcher is ours, and it uses something Shopify already publishes: each page knows its own equivalent on the other domain, translated URL included. So the switch always lands on the same page in the other language, there is no list of URL pairs for anyone to maintain, and nothing needs re-running when the catalogue changes.
Shipping that knows what the thing actually is
A surfboard is not a parcel. The old shop offered one shipping setup for everything, which produces orders that cannot be fulfilled and a phone call to explain why.
We classified the catalogue by what each item physically is: 16 pieces are too long for any courier and are pickup-only, 84 will not fit a parcel locker but travel fine by courier, the rest are ordinary parcels. Shoppers now see only the delivery options their order can actually use, and nobody checks it by hand before fulfilment.
The same principle runs through the build. What a shop genuinely needs from a service - a carrier, a payment provider - is an app. What is really just configuration and a script is ours, so the store is not carrying a monthly bill for the lifetime of the business.
The funding paperwork too
The project ran under a state digitalisation programme, which wanted formal evidence of what changed: the old system against the new, scored across ten criteria, and the working time it gives back - around 96 hours a year. We wrote it and handed it over ready to submit.
Stack: Shopify, Admin GraphQL API, Liquid, Python. Two markets, two domains, two inventory locations.