HjulMatch.se – A self-built marketplace that aggregates motorhome and caravan listings from multiple Swedish dealers every night – searchable in ways no single dealer site can match.
Project Snapshot
Role: Founder, Product Owner, Concept Development & AI Coordination
Project: HjulMatch (hjulmatch.se)
Audience: Motorhome and caravan buyers in Sweden
Scope: Marketplace/search engine developed with Claude Code that includes an automatic scraping pipeline, data modelling, affiliate integration, SEO content, and video production
The Challenge
When I was looking for my next motorhome, I kept hitting the same wall: no dealer site let me filter by the things that actually mattered to me. I wanted to search by a specific bed type, a certain heating system, or a compressor fridge – details that matter enormously once you’ve actually owned a motorhome, but that no single dealer treats as a real filter. Every dealer publishes listings on its own site, with its own structure and level of detail, so finding a specific combination meant checking dealer after dealer by hand.
That personal frustration became the starting point for HjulMatch. The goal is simple: to be the smartest way to find the right motorhome or caravan – with filters that actually matter to you as a buyer, like brand, bed type, heating system, and size. The ambition is for HjulMatch to become Sweden’s largest aggregator of motorhomes and caravans.
My Approach
Turning that personal need into a real product meant building HjulMatch as an aggregating search engine rather than a listings platform – the service creates no listings of its own, but scrapes, normalises, and structures data from dealers’ existing websites.
Three things ran in parallel: a lean, cost-efficient technical foundation (Next.js, Supabase, GitHub Actions – deliberately without Docker, Redis, or a dedicated VPS until there’s an actual need for them); a data model that makes vehicles genuinely comparable even though every source structures its data differently; and a branding and content effort that makes the service discoverable and trustworthy from day one.
As a solo operator who isn’t primarily a developer, I coordinate all code implementation through separate AI-driven development sessions (Claude Code) – one for frontend/design, one for backend/scraping – while owning product decisions, documentation, and quality assurance myself.
What I Delivered
- Concept development and branding: name, positioning, logo, colour palette, and tone of voice (“Your Home. Your Adventure. Your Match.”)
- Technical architecture: Next.js frontend on Vercel, Supabase (PostgreSQL) as the data layer, TypeScript-based scraper pipeline
- Nightly scraping pipeline via GitHub Actions with parallelised jobs, normalising vehicle data from 15+ dealer sources
- Advanced filtering on attributes such as bed type, chassis, fridge type, and heating system, with full-text search in progress
- Affiliate integration (Adtraction) and a standalone advertising system for an independent revenue stream
- SEO-driven content strategy: 15+ guide articles targeting buyers’ actual search behaviour
- Video pipeline: an automatic Canva-generated explainer video for every guide article, published on a dedicated YouTube channel
- Review-matching feature surfacing external reviews linked to each vehicle model
- GA4 tracking via GTM to follow user behaviour and conversion
Business Impact
- A working marketplace in production, built and operated by one person, without unnecessary infrastructure cost
- 15+ active scraper sources running fully automated every night, with no manual handling
- Independent affiliate revenue and advertising system, live and decoupled from third-party platforms
- SEO content that positions HjulMatch as a knowledge source, not just a search tool
- A scalable foundation where new data sources, filters, and features can be added without rebuilding the architecture
- Concrete proof of the ability to drive a product end-to-end: concept, brand, technical architecture, operations, and content
Tools & Technology
- Frontend: Next.js, Vercel
- Data: Supabase (PostgreSQL), Postgres full-text search, pg_trgm
- Scraping: Node.js/TypeScript, Playwright, one module per source
- Automation: GitHub Actions (nightly, parallelised jobs)
- AI: Claude API for attribute extraction from listing text, Claude Code for development work
- Affiliate & advertising: Adtraction, Amazon, own advertising system
- Content & video: Canva Pro (video and image production)
- Analytics: GA4 via Google Tag Manager
Continuous Development
HjulMatch.se is an ongoing project. Next steps include attribute-based landing pages (by bed type, for instance), distance-based search from the user’s location, and additional data sources to further broaden the listings available.