One automotive parts retailer had 500,000 listings online and a 10% return rate. A year after fixing its fitment data, returns sat at around 3%. No new ad budget, no site rebuild, just accurate answers to the one question every car parts buyer asks, will this fit my vehicle?
That’s the story Radoslav (Rado) Chernaev, CEO of Autofixa, shared on the eCom@One podcast. Autofixa builds eCommerce infrastructure for automotive brands, retailers, distributors and dealers, from major OEMs to smaller independents, and Rado has spent 15 to 20 years selling online. Here are the lessons that stood out.
1. Car parts weren’t built for eCommerce
Fashion and tech are close to tailor-made for online selling. A trainer has a size, a colour and a set of photos. A laptop has a spec sheet. Shoppers compare and buy.
Automotive is a different animal. Retailers manage catalogues of 90,000 to 200,000 parts, often running on legacy systems with thin product data and few or no images, because photographing every part is expensive. On top of that sits fitment: a single wiper blade might fit hundreds of combinations of make, model, year, engine and chassis. In most categories the buyer asks “do I want this?” In automotive, the first question is “does it fit my car?”, and the retailer has to answer it accurately across the whole catalogue.
2. Bad data becomes a problem at scale
When the retailer with 500,000 listings came to Autofixa, the cause of its returns was easy to spot. A large share of listings had no fitment at all, and what fitment did exist was around four years out of date. Customer service was answering compatibility questions by hand, which slowed conversion and pushed impatient buyers to competitors who let them check against their vehicle instantly.
Rado’s team applied the Pareto principle: 20% of products were driving 80% of returns. Fixing that 20% first brought the return rate down quickly, and it settled around 3% within a year. Marketplace rankings improved too.
The knock-on effects of getting fitment wrong go well beyond the cost of a return. Customer service time climbs, reviews suffer, and reviews increasingly feed into how Google and AI assistants like ChatGPT and Claude decide which brands to surface. On low-margin parts, a single return can wipe out the profit entirely. Fitment isn’t a one-off upload either. It needs updating constantly as new vehicles and parts appear.
3. AI helps data teams scale, it doesn’t replace them
Autofixa sources fitment from specialist catalogue and data providers, then uses its own automation and, over the last 18 months, AI to process it faster. But every output is still checked in-house. A hallucinated fitment means the wrong part arriving on someone’s driveway, repeated across thousands of listings. Asked whether AI will replace product data teams within three years, Rado’s answer was firm: no, it will help them scale, because the category is still too complex.
4. Buyers are shopping on price and age of vehicle
Several years of pressure on household budgets, combined with people keeping their cars for longer, means buyers are making very different choices depending on what they drive. Rado’s own example: wiper blades for his two-year-old Kia Sorento ranged from £35 for the original part to £30 for premium aftermarket, around £18 to £20 for standard options, and as little as £2. He chose the premium aftermarket. On a 15-year-old car, he’d have gone cheaper.
Buyers now choose between original equipment, a wide spread of aftermarket brands and green (refurbished) parts. That makes education part of the retailer’s job. Listing 20 similar products isn’t enough. You need to explain why one costs £10 and another £40, whether the difference is warranty, longevity, specification or brand.
5. Marketplaces give brands more control, not less
The question Autofixa hears most from big OEM brands is whether selling on Amazon or eBay will dilute the brand. Rado’s view is that the opposite is true. Most of a brand’s catalogue is probably already on marketplaces, listed by resellers who are there to make a sale, not to represent the brand, and often with poor or wrong fitment. When a buyer gets the wrong part, it’s the brand that takes the hit.
Selling directly lets a brand own the listing, the data and the customer experience, while tapping into enormous existing traffic. For a brand starting out, his advice was marketplace first, every time, with eBay ahead of Amazon for auto parts today.
What comes next
Rado expects several shifts at once: more brands going direct to consumer for the margin, continued marketplace growth as AutoDoc, B&Q and Temu move into parts alongside eBay and Amazon, and AI driving better fitment technology. The common thread is that the brands with the cleanest, most current data will be the ones customers, marketplaces and search engines trust.
Hear the full conversation
The full episode also covers what a two-year OEM tender really involves, from security and GDPR to financial due diligence, plus Richard’s dilemma over a carbon CSL splitter for his BMW E46. If fitment, returns or marketplace strategy is on your mind, it’s well worth a listen.
Watch the full episode here
