Auto One had genuine retail strength and a customer who knew what car they drove but not what part number they needed. The automotive aftermarket is a fitment problem before it is a commerce problem: the same component varies by make, model, year, engine variant and sometimes build month, and a plausible-looking wrong part generates a return, a refund and an eroded reputation. Extending a strong physical network online meant the site had to be at least as reliable as asking someone behind the counter — while also being fast on a phone, because a meaningful share of the category is bought standing next to the vehicle.
We built on Shopify Plus with the aftermarket’s actual logic layered over it: VIN-based search and advanced fitment filtering, so a customer converges on parts confirmed to fit rather than parts that merely match a keyword. Inventory connects through API so availability reflects reality. The interface was designed mobile-first for one-handed use in a driveway or a workshop, and deployed as a PWA for speed on poor connections. Retargeting runs on dynamic remarketing and programmatic networks keyed to the specific vehicle and part a customer viewed — in a considered category, relevance is what recovers an abandoned cart, not frequency.
Shopify Plus with custom Liquid · VIN-based search and advanced fitment filtering · inventory API integration · PWA deployment for mobile speed · dynamic remarketing and programmatic retargeting · Performance Max and Meta Advantage+ on a governed feed · GA4 dashboards through to transaction.
Performance Max and Meta Advantage+ run against a governed product feed with automated exclusion of out-of-stock and low-margin lines. Cart-abandonment sequences are triggered on behavioural signals rather than a fixed timer, and creative is matched to the exact vehicle context the customer was browsing. Search-term clustering surfaces the fitment queries the catalogue cannot yet answer, which is the most reliable roadmap for what to build next.
Conversational fitment — registration or VIN in, confirmed-compatible parts out — grounded strictly in the fitment database, with an explicit refusal where the data cannot confirm a match rather than a confident guess.