This work was shipped by the team behind StoreSmith, before StoreSmith existed. The merchant keeps its name. Figures are rounded on purpose.
the problem
A multi-brand watch-and-jewellery retailer sold entirely through physical stores. Its customers bought expensive pieces, often after several visits, and usually from a sales associate they knew. The business wanted an online channel without breaking any of that.
Stated the way the merchant put it: we need to sell four-figure pieces online, show only what is really in a store, keep every brand’s rules, and make sure the customer record our associates rely on stays complete.
what we built
Store inventory as the online catalogue. There was no separate online stock. Inventory came from the store systems, synced into Shopify Plus by location, so the site only offered pieces that physically existed and could say where. Reservations and sales flowed back the other way through webhooks, with idempotent handling so a retried event never sold the same piece twice.
One customer record across channels. Online customers, orders and enquiries synced to the CRM the associates already used. An associate could see that a client had browsed or bought online; a client could buy online and collect in store from someone who knew the order.
An authentication trust layer. High-ticket buyers need to know a piece is genuine before they pay. Product pages and the checkout carried authenticity, warranty and provenance information per brand, driven from structured product data rather than hand-written copy. Checkout UI extensions put the right assurances in front of the buyer at the moment of payment.
A checkout built for high-ticket orders. Payment, delivery and fraud handling were configured for order values well above typical fashion. Shopify Functions handled the brand-specific rules: which pieces could ship and which were collect-only, and which payment options applied.
New brands as configuration, not projects. Each brand’s rules, content and assurance blocks lived in metaobjects and settings. Adding a brand meant data entry and a review, not a release, which is why a new brand could go live in days.
how it shipped
We launched with a subset of brands and stores, watched real orders end to end, and then widened. Inventory sync ran in read-only mode against the stores before it was trusted to drive the catalogue. Associates were part of testing, because the CRM view was as important to them as the storefront was to the buyer.
results
- Online AOV 2–4× the category norm.
- Four-figure AOV as the normal case, not the exception.
- New brands live in days once the pattern was in place.
what we’d do the same again
Keep the store as the source of truth. Make brand onboarding a data task. Put the trust signals where the money changes hands, not three pages earlier.
what we’d do differently now
The inventory connector, the CRM mapping and the per-brand rule functions were hand-written and hand-tested. Today the loop drafts them against real payloads, and a senior engineer reviews and signs every merge. The decisions stay human; the typing does not.
stack
Shopify Plus · checkout UI extensions · Shopify Functions · metaobjects · location-based inventory sync · CRM sync · webhooks with idempotent handling.
Connecting a CRM or store system? See apps, automation or crm connect.