JJudgejudge.app →
Score

Vercel × Customer-Centric Judge

5/8/2026, 8:00:46 AMEdited by gallery-seed

Who is judging this
Rubric

Looks at the change from the perspective of the end customer.

Their top concerns (weight share)
  • Clarity
    29%
  • Trust
    27%
  • Empathy
    22%
The score below reflects how this specific rubric weighs the page. If their priorities don't match your real audience, recalibrate before acting on the recommendations.
54.9/ 100First
What changed since last time
Score: 54.9 (first iteration — no prior baseline).
Why the judge scored it this way— show full

Vercel's homepage is technically dense and clearly written for developers who already know terms like 'Fluid Compute,' 'multi-tenant,' and 'Framework-Defined Infrastructure.' A non-technical user — a founder, a marketer, a design lead — encounters a wall of product labels with no plain-English context to anchor them. Social proof stats are compelling but unattributed, reducing their persuasive weight. The hero CTAs are well-placed, but the page quickly buries them under a broad product catalogue and a code snippet, slowing the non-technical visitor's journey to a decision. Trust signals like the model leaderboard exist but lack sourcing, which a skeptical buyer will notice.

Vercel's homepage is technically dense and clearly written for developers who already know terms like 'Fluid Compute,' 'multi-tenant,' and 'Framework-Defined Infrastructure.' A non-technical user — a founder, a marketer, a design lead — encounters a wall of product labels with no plain-English context to anchor them. Social proof stats are compelling but unattributed, reducing their persuasive weight. The hero CTAs are well-placed, but the page quickly buries them under a broad product catalogue and a code snippet, slowing the non-technical visitor's journey to a decision. Trust signals like the model leaderboard exist but lack sourcing, which a skeptical buyer will notice.

What to fix first · ranked by impact = how far below “Good” × weight. Fix #1 first.
  • 1
    Clarity
    Weakweight 29%score 5.0/10

    Plain language, no internal jargon.

    DiagnosisHeavy jargon leaks throughout — 'Fluid Compute,' 'Framework-Defined Infrastructure,' 'Active CPU pricing,' 'BotID,' 'Composable Commerce,' and 'multi-tenant platforms' are used without plain-language explanation, leaving non-technical users without a mental model of what they're actually buying.

    Do this nextAdd a one-line plain-English gloss beneath each product name in the nav and hero sections (e.g., 'Fluid Compute — pay only when your server is actively working') so first-time visitors can self-qualify without clicking away.
  • 2
    Empathy
    Weakweight 22%score 5.0/10

    Anticipates user concerns and edge cases.

    DiagnosisSocial proof metrics ('build times went from 7m to 40s,' '95% reduction in page load times') are presented without attributing them to named customers or use cases, so a non-technical buyer can't tell whether the claims apply to their situation.

    Do this nextAttach each stat to a named customer and their context (e.g., 'Acme's marketing site saw a 95% reduction in page load times after migrating from AWS') to let visitors see themselves in the story and trust the relevance of the numbers.
  • 3
    Trust
    Fairweight 27%score 6.0/10

    Honest, non-misleading messaging.

    DiagnosisThe 'Top models on May 8, 2026' leaderboard with specific usage percentages (e.g., 'Gemini 3 Flash 25.2%') is a credibility asset, but the source and methodology are never explained, which undermines rather than builds confidence.

    Do this nextAdd a tooltip or footnote to the model leaderboard stating the data source and date range (e.g., 'Based on Vercel AI Gateway requests, rolling 30-day average') so the figures feel auditable rather than arbitrary.
  • 4
    Speed to value
    Fairweight 22%score 6.0/10

    User reaches benefit quickly.

    DiagnosisThere are two clear CTAs ('Start Deploying' and 'Get a Demo') early in the hero, which is good, but the page then expands into five distinct product categories, multiple SDK code snippets, and a 10-item model leaderboard before re-surfacing a CTA, slowing a non-technical visitor's path to understanding what to do first.

    Do this nextInsert a persistent, sticky CTA bar or a guided 'What are you building?' selector (AI App / Web App / Ecommerce) near the top of the page to route visitors to the most relevant template or demo in one click rather than asking them to scroll through the full product surface.

Metrics

Looks at the change from the perspective of the end customer.

  • trust
    60%
  • speed to value
    60%
  • clarity
    50%
  • empathy
    50%
Judged by Sonnet 4.6 · customer-centricPublic link · read-only

Want quality scoring on your own code? Try Judge.