Case study · 2025–2026
Connect Car Parts: A.B.S. brake parts on Amazon DE, eBay DE and Magento
How one operator runs a multi-channel parts business with structured product data, feed automation and AI-assisted content, with a human on pricing, brand voice and compliance.
~400
Active SKUs live
2,400+
Product records with generated content
Magento · Amazon DE · eBay DE
Channels
every 30 min
Order monitoring
Context and role
Connect Car Parts is the e-commerce operation of ABS All Brake Systems: a Dutch seller of brake parts (discs, pads, hoses, wheel-bearing kits) under the A.B.S. brand and complementary brands such as Frankberg. Sales run through its own Magento storefront and on Amazon DE and eBay DE; Bol.com and the full A.B.S. assortment on eBay DE are being rolled out. I've worked here as e-commerce manager since December 2025 and run the marketplace side end to end.
Two numbers that are often confused: the active catalogue that is live counts roughly 400 SKUs with vehicle-fitment data (K-types) and OE cross-references; the number of product records that have gone through the content pipeline exceeds 2,400, because every record gets its own version per country and channel and because the wider A.B.S. assortment is being prepared.
The challenge
A brake part only sells when the buyer is certain it fits their car. That requires fitment data, OE numbers and technical specifications per SKU, plus different title, attribute and compliance rules per marketplace (GPSR, ISO, brand mention). Done by hand, a publish-ready listing took 45 minutes on average; with thousands of records across three countries and three languages that was unsustainable, and errors in fitment or claims led straight to returns or rejections.
The approach
- Structured product data import from ERP/PIM and the A.B.S. TecDoc export, with K-type mapping per vehicle
- Feed management through Channable with brand-split GPSR and ISO rules per channel (Amazon DE, eBay DE, Bol.com)
- AI-assisted content generation: titles, bullets, A+ Content and backend keywords per country, with source and confidence labels
- Automatic policy validation (Amazon rules, byte limits, forbidden claims) before publishing; human review on pricing and brand voice
- Quality score per listing with prioritisation on revenue opportunity; visual content such as brand infographics and comparison tables
- Automated VIN-based parts lookup in the storefront, a first in the industry, so buyers find the right part by licence plate
- Order monitoring every 30 minutes through n8n with failure alerts; daily radar on Channable feed quality and marketplace rules
What it delivered
- ✓Average quality score of 94/100 across all generated listings (internal measurement of the content pipeline, 2025)
- ✓Content creation time from 45 minutes to 3 minutes per SKU (−93%, internal time measurement)
- ✓Zero policy rejections after implementing the validation step
- ✓Consistent brand messaging across 3 countries and 3 languages
- ✓eBay, Amazon and Bol.com connected into one marketplace operation with a single product-data source
- ✓Revenue forecasts and KPI reporting used weekly by management
Ongoing work
The Amazon DE catalogue is being launched including German VAT registration, the full A.B.S. assortment is moving to eBay DE and Bol.com with marketplace-specific content rules and vehicle compatibility files, and a weekly competitor scan tracks price and assortment movements on Amazon DE and eBay DE. What I keep learning about agents reporting success without evidence is documented in my articles on AI reliability and meta-monitoring.
AI e-commerce automation · Amazon services · Article: my AI agent invented success (Dutch)
Deliverables
The same for your catalogue?
This approach works for any large catalogue with technical product data. See the service pages or book an intro call.
Book an intro call