Bike Crash Claims connects motorcyclists who have had a non-fault accident with regulated partner firms who pursue their compensation, arrange repairs and can get them back on a replacement bike. It is an introducer, not a law firm and not regulated itself, which makes one thing non-negotiable: everything the site says has to be accurate. We rebuilt it to be fast, found, and honest.
The challenge
The original site was a Hostinger Horizons React single-page app. It rendered in the browser rather than on the server, which is a poor start for search, the content was thin, and some of the structured data overreached, claiming to be a local business in eleven different cities and faking a “last updated” date on every page. In a regulated, trust-sensitive space, that is exactly the kind of thing you cannot ship. The brief was to be genuinely findable and to convert well, without a single overclaim.
A static rebuild, engineered for search
We rebuilt the site as an exact replica on Astro, static-first, with React used only for the few genuinely interactive pieces. Then we grew it into a real content library: claim-type pages, richly written accident-scenario articles, 22 location pages and 21 guides, over 80 pages in all. Every image was re-encoded to WebP, cutting image weight from 48MB to 4MB so pages load fast on a phone at the roadside. That is the foundation of web development that ranks.
Honest schema, the part most agencies skip
The most valuable work here was subtraction. We tore out the false “local business in eleven cities” markup and replaced it with honest Service structured data that simply states the areas served. We removed the faked freshness date. We consolidated a messy entity graph into one clean, correct Organization. Then we layered the schema that genuinely helps, Article, FAQPage, HowTo and Breadcrumb. Accurate structured data and SEO is not just good practice here, it is the whole reputation of an introducer in a regulated market.
Built to be cited by AI
Riders increasingly ask an AI assistant before they ask anyone else. So we made the site answerable: an llms.txt that discloses exactly what the business is (and isn’t), AI crawlers explicitly welcomed, and direct-answer content that models can quote. The location pages even carry bespoke coverage maps drawn from real Ordnance-Survey-style boundary data as inline SVG, so nothing calls a third-party map service before the visitor has consented. That blend of AI-ready content and automation is how you get surfaced in the answers people now trust.
A conversion path built for a regulated space
Enquiries run through a clean three-step form protected by Turnstile and a honeypot, with consent captured by submission, the exact wording shown to the rider is stored with a timestamp against their enquiry. A jurisdiction step routes England and Wales enquiries to the regulated partner and keeps others on-site, and the whole funnel reports to Meta’s Conversions API server-side with browser and server events de-duplicated. Google Consent Mode v2 means nothing fires before the visitor agrees. It is conversion-focused design that respects the rules and the reader.
The result
A fast, honest, 80-plus-page site, live at bikecrashclaims.co.uk, built to be found by riders and by the AI tools they ask, and to connect them with the right regulated firm, the compliant way. It is the standard we bring to trust-sensitive professional services: be accurate, be found, and never overclaim.