45x
More Text For AI Crawlers
12
Pages, Each With Its Own Canonical
81%
Lighter Hero Image
AI
Search-Ready
THE PROBLEM
Levr had a fast, well-designed vehicle finance site that search engines could barely see. Every page returned the same title, the same description and no canonical, so Google could not tell them apart — and Search Console showed the result: two pages indexed out of eleven, several never crawled at all. Worse, because the page was assembled in the browser, the bytes sent to anything that does not run JavaScript contained no words. Fetched as GPTBot, a landing page returned 30 characters and no heading. A site that cannot be quoted cannot be cited, and citation is the whole of AI search.
OUR APPROACH
We diagnosed it with live Search Console data rather than assumptions, then fixed the cause. A build step now renders every page and writes the result into the HTML, so ChatGPT, Claude, Perplexity and Bing receive the same words a visitor sees — 1,360 characters and a real heading where there had been 30 and none. Every URL gained its own title, description and canonical. We added llms.txt, a working favicon, six security headers where there had been one, and cut the hero image by 81%. Two new pages target the highest-demand, lowest-competition searches in NZ car finance. The whole thing is pinned by 124 browser tests, and the build fails rather than shipping a page with no words in it.
SEE IT LIVE
Car loan repayment calculator