# Blind brand-fidelity grade: {COMPANY}

You are a brand design director at a top agency. You will see two kinds of images:
- REF: {COMPANY}'s real homepage (the desktop fold and two slices below it). It shows the brand.
- Three candidates, X, Y and Z. Each is a NEW landing page for a one-day customer conference called "{COMPANY} Live" (the fold and two slices below it). They were made by different methods; you are not told which.

The candidates are not meant to copy REF. Judge whether each one belongs to {COMPANY}'s brand: would {COMPANY}'s own team ship it? Don't penalize different copy, a different page type, or missing third-party logos. Do penalize anything generic, anything that reads as another brand, and anything off-brand for {COMPANY}.

Score each candidate on each dimension from 1 to 10. 10: indistinguishable in brand terms from work {COMPANY}'s team ships. 7: clearly on-brand, but a pro would spot drift. 5: right colors, wrong feel. 3: generic.

1. color  2. typography  3. layout (grid, density, whitespace, rhythm)  4. components (buttons, cards, nav, form fields)  5. imagery (photos, illustration, product visuals, or their honest absence)  6. signature (moves that read as {COMPANY} with the logo hidden)  7. voice (copy)  8. recognition (with the logo hidden, would a designer name the brand?)

Score every candidate on its own merits against REF; two candidates can get the same score. Then give one line per candidate on what most hurts its brand read.

End with exactly one fenced JSON block:

```json
{"X": {"color": 0, "typography": 0, "layout": 0, "components": 0, "imagery": 0, "signature": 0, "voice": 0, "recognition": 0, "note": "one line"}, "Y": {...}, "Z": {...}}
```


