Relark.dk: From invisible to found by AI
How an AI company — built by the same founder as Geoa — got its own AI visibility under control.
Case Study
Company
Relark — A psychological copilot for Nordic middle managers. Big Five-based leadership sparring and conversation scripts for difficult talks.
Industry: AI SaaS, Leadership
Size: Startup (1-5 employees)
Audience: Middle managers with 5-25 staff in DK/NO/SE
The challenge
Relark is a modern AI SaaS — built as a React SPA (Single Page Application). Content is rendered in the browser, not on the server. AI crawlers like GPTBot, ClaudeBot and PerplexityBot saw essentially just an empty page.
Even though Relark had implemented llms.txt and JSON-LD, the lack of SSR and AI crawler access blocked AI visibility. GEO Score: 50.3.
What Geoa found
AI crawlers blocked
robots.txt did not allow GPTBot, ClaudeBot, PerplexityBot or Google-Extended — the AIs could not crawl the site.
Client-side rendering
React SPA without SSR — AI crawlers saw an empty div. Content was only available via JavaScript.
Weak E-E-A-T profile
Missing author bios, external mentions and LinkedIn presence = low AI authority.
The solution — Auto Fix via Geoa
AI crawler access opened
robots.txt updated with Allow for GPTBot, ClaudeBot, PerplexityBot, Google-Extended and OAI-SearchBot — so the AIs can actually crawl the content.
Server-Side Rendering
Prerendering of critical content, so AI crawlers see content instead of an empty JavaScript container.
E-E-A-T strengthened
Author bio with credentials (MPA, officer in the Armed Forces), LinkedIn link, and an external-mention strategy.
Content structure optimised
H2/H3 headings, tables with prices/features, and shorter paragraphs (40-60 words) for better AI extraction.
Results
GEO Score before — grade F (9 July 2026)
GEO Score after — grade C (20 July 2026)
From F to C — technical visibility, not promised customers
Top 5 fixes that make the biggest difference:
- AI Crawler Accessibility (score: 0 → 100) — Open to GPTBot, ClaudeBot, PerplexityBot
- Structured Data (score: 0 → 100) — Fix JSON-LD into valid blocks
- Internal Links (score: 20 → 100) — Add 5+ internal links with descriptive anchor text
- Server-Side Rendering (score: 30 → 100) — Prerender critical content
- E-E-A-T Content Quality (score: 38 → 100) — LinkedIn, G2, author bios
“Relark had both llms.txt and JSON-LD in place and still scored an F. That is the point: the individual signals mean nothing if the crawler never reaches the page.”
What was used
- GEO Score across 15 dimensions
- Auto Fix
- AI crawler access
- Server-side rendering
- Weekly re-measurement
Related cases
AddonNordic.dk: From F to B in 20 days
We ran Geoa on our own company domain and recorded every single measurement — including the one that went the wrong way.
Geoa.app: From B to A in 45 days
A site built by people who make their living from AI visibility ought to have nothing left to fix. It did. 11 measurements later, the score was 5.3 points higher.
Cases: what Geoa found, and what the numbers did afterwards →
See for yourself — get your own GEO Score
Relark.dk was a real case with real data. Run the same test on your own website — free, no sign-up.