Your landing page’s weakest score is Trust.
How Fulcrum ranks
Trust
48PoorProof that the product is real, competent, and safe to try.





| Trust read | |
|---|---|
| Fulcrum | "Highlighted Work" benchmark chart supporting "Fable is SOTA at CIFAR Speedrun" |
| "Non-public information and model access provided by:" with Anthropic, Google DeepMind, Meta, and OpenAI logos on the Fr | |
| "OUR PARTNERS" logo strip naming AI Security Institute, Microsoft, Meta, Anthropic, OpenAI, Amazon, NIST, DeepMind, and | |
| "Trusted by leading publications and institutions" logo strip (Our World in Data, DSIT, Government of the Netherlands, N | |
| "Our Work" cards showing real research output, including the essay "Some Focus Areas for Embedded Evaluations and How to |
CTA
58FairCopy, size, verb, cost-feel — does the primary CTA earn the click?




Problems
- small, muted button lacks visual weight
- 'updates' is vague — no named payoff
- sits below nav links, easy to miss
- no supporting microcopy on the form
Value props
58FairTitle, subtitle, hero — do they sell the product's true value?
Fulcrum
We work on measuring and optimizing model performance in fuzzy domains.
Working
Could improve
Information hierarchy
62FairDoes the design help or inhibit what the page is trying to say?

What works
What to improve
Mobile web
68FairThe phone experience as a first-class citizen.

What works
What to improve
