VelvetYC F25 landing page report.
← Back to YC F25

Your landing page’s weakest score is Trust.

Fix Trust

Score over time

90807060504030Scale AI 62 on May 30 (Wayback)Scale AI 62 on Jun 15 (Wayback)Scale AI 49 on Jul 11 (Wayback)Scale AI 64 on Aug 23 (Wayback)Scale AI 62 on Sep 5 (Wayback)Scale AI 61 on Sep 20 (Live)Surge AI 46 on Jun 9 (Wayback)Surge AI 45 on Jul 15 (Wayback)Surge AI 44 on Jul 24 (Wayback)Surge AI 46 on Aug 13 (Wayback)Surge AI 44 on Aug 21 (Wayback)Surge AI 46 on Aug 31 (Wayback)Surge AI 46 on Sep 20 (Live)Encord 71 on Jun 14 (Wayback)Encord 72 on Jun 23 (Wayback)Encord 71 on Jul 6 (Wayback)Encord 71 on Jul 18 (Wayback)Encord 70 on Jul 30 (Wayback)Encord 73 on Aug 23 (Wayback)Encord 74 on Sep 20 (Live)Mercor 75 on Jun 25 (Wayback)Mercor 76 on Jul 14 (Wayback)Mercor 77 on Jul 23 (Wayback)Mercor 74 on Aug 12 (Wayback)Mercor 59 on Aug 28 (Wayback)Mercor 57 on Sep 5 (Wayback)Mercor 54 on Sep 20 (Live)Velvet 45 on May 31 (Wayback)Velvet 54 on Sep 22 (Live)May 31Sep 22
Past reportsShow past reportsHide past reports
  1. Sep 22, 202654LiveOverall +9 · Value props +34 · Information hierarchy +10Viewing
  2. May 31, 202645First tracked report — strongest Information hierarchy 52, weakest Value props 34View →

Trust

38Poor

Proof that the product is real, competent, and safe to try.

Encord
Trust read
VelvetNo visible trust signal identified.
Scale AIimplicit context Scale AI may benefit from pre-existing familiarity among its target audience.
Surge AI"New Frontier data and RL environments, off the shelf" announcement banner
EncordCustomer logo strip below the hero: Path Robotics, Woven by Toyota, Zipline, Pickle Robot, Toyota Research Institute, We
Mercorimplicit context Mercor may benefit from pre-existing familiarity among its target audience.

CTA

48Poor

Copy, size, verb, cost-feel — does the primary CTA earn the click?

we'd love to work with you
Velvet
Encord
Scale AI
Mercor

Problems

  • no hero button or input
  • inline text link, tiny footprint
  • no action verb, 7 words
  • nav 'Sign Up' mismatches intent

Mobile web

54Fair

The phone experience as a first-class citizen.

Mobile web annotations on the mobile capture1234
  1. Primary CTA below fold — The only action, 'we'd love to work with Velvet', is an underlined text link at the very bottom of the fold, barely surviving as a tappable target.
  2. Uncollapsed nav links — 'Sign Up' and 'Sign In' sit as full text links beside a hamburger, a double nav pattern that clutters the header on a phone.
  3. Dense body paragraphs — Three desktop-length paragraphs fill the entire viewport, stacking competing messages instead of one statement and one action.
  4. Text-link CTA styling — The lone conversion element is styled as inline underlined text, not a thumb-friendly full-width button.

Information hierarchy

62Fair

Does the design help or inhibit what the page is trying to say?

Information hierarchy annotations on the desktop capture123
  1. Text-link primary CTA — "we'd love to work with Velvet" is an underlined sentence fragment at the fold's bottom edge, not a button — it fails the one-second find test.
  2. Third paragraph block — Three same-weight paragraphs stack into a text wall; the key claim about spatial reasoning is buried mid-block instead of pulled out as a beat.
  3. Sign Up / Sign In — Two equally weighted account links are the only button-like elements in the fold, outranking the actual conversion path below.

Value props

68Fair

Title, subtitle, hero — do they sell the product's true value?

Velvet’s headline sellsVelvet helps train world modelsVelvet is a product and research company focused on world models.

Working

  • Honest, jargon-correct claim for the ICP
  • Sharp problem framing: spatial reasoning gap
  • Clear audience: AI labs and consumer companies

Could improve

  • Zero numbers or proof points anywhere
  • Subhead restates the headline, adds little
  • Hero art decorates, doesn't demonstrate

Competitors

inferred from product, audience, and value prop