Your landing page’s weakest score is CTA.
How WebArena ranks
CTA
42PoorCopy, size, verb, cost-feel — does the primary CTA earn the click?



Problems
- no button or input anywhere
- cards use bare arrow glyphs only
- arrows never say where they lead
- no verb on any clickable element
Value props
48PoorTitle, subtitle, hero — do they sell the product's true value?
Projects
A suite of benchmarks for building autonomous web agents.
Working
- Subhead names the real category clearly
- Venue proof: NeurIPS 2024 Oral, ICML 2025
- Card blurbs are concrete, not buzzwords
Could improve
- 'Projects' H1 sells nothing
- No numbers: tasks, sites, environments
- No outcome or emotional hook stated
Trust
55FairProof that the product is real, competent, and safe to try.

How WebArena compares
Trust read
Information hierarchy
58FairDoes the design help or inhibit what the page is trying to say?

What works
What to improve
Mobile web
62FairThe phone experience as a first-class citizen.

What works
What to improve
