Batteries are a search-led purchase with an unforgiving catalogue. A customer arrives knowing a device, a vehicle or a part code — rarely a product name — and they need the one item that fits. Thousands of SKUs across automotive, industrial, marine, mobility and consumer categories, many differing by a single specification, meant the wrong result was worse than no result: a mis-sold battery becomes a return, a refund and a lost customer. Stock moved constantly across the store network, so anything the site claimed about availability had to be true at the moment of the claim, not at the last overnight sync.
We rebuilt the catalogue around how people actually search: fitment and application first, product name last. Categorisation was restructured so a customer can arrive by device, vehicle, capacity or terminal type and converge on the correct product, with specification filters that eliminate incompatible items rather than merely sorting them. Inventory syncs with the ERP in real time so availability is accurate at the point of decision. Product data was made machine-readable with full product schema, which serves both Google Shopping surfaces and the AI answer engines increasingly fielding “what battery fits” questions before a customer reaches any website. Paid media runs on that same clean product feed, so a campaign can never advertise something the warehouse cannot ship.
High-volume eCommerce with fitment and application-led categorisation · specification filtering that excludes incompatible SKUs · real-time ERP inventory synchronisation · secure payment gateways · product and offer schema across the catalogue · Performance Max on a governed product feed · GA4 commerce tracking through to transaction.
Performance Max and feed-driven bidding operate against the live product feed with automated rules that suppress out-of-stock and low-margin lines, so budget follows what is genuinely sellable. Search-term analysis is clustered automatically to surface fitment questions the catalogue does not yet answer — each one is a content gap with revenue attached. Structured product data is maintained specifically so answer engines can quote correct specifications rather than guess.
Guided fitment as a conversational step — device or vehicle in, correct product out, grounded strictly in the product database with a refusal when the data cannot support a confident match. In this category a wrong answer costs a return, so the boundary matters more than the coverage.