# Twenty-brand independent presentation benchmark

Two judges independently inspect every readable rendered mobile390x844 and desktop1440x900 frame for each selected PDP. Score the sampled page, not the whole brand. Do not use reputation, sales or unverified conversion assumptions. Full capture manifest required; blocked or materially incomplete evidence receives null overall and reasons. Scores are subjective visual/presentation judgments, not measured CVR, product validation or release certification. Compare category-appropriate needs; more sections or UGC is not automatically better. Use tools to VIEW every screenshot, not source HTML or manifests alone.

## Fixed dimensions and weights

| Key | Dimension | Weight | Judge |
| --- | --- | ---: | --- |
| craft | Visual craft | 25 | Composition, typography, hierarchy, spacing, image placement, finish |
| identity | Brand identity | 20 | Ownable visual language, warmth, consistency; distinguish polished familiarity from an authored brand |
| journey | Page flow | 20 | Sequencing, pace, density, readable progression, decision timing |
| trust | Trust presentation | 15 | Genuine reviews/UGC, accessible evidence, candor, reassurance, proof placement |
| offer | Offer clarity | 10 | Visible options, price/shipping/refill terms, chosen state, CTA clarity |
| mobile | Mobile presentation | 10 | Full 390px journey, readability, control arrangement, content priority; supported section checks at other widths |

Anchors: 0–2.9 unusable/misleading; 3–4.9 major failures; 5–5.9 functional but weak or generic; 6–6.9 competent with meaningful gaps; 7–7.9 strong with identifiable weaknesses; 8–8.9 excellent with limited material gaps; 9–9.4 exceptional across the assessed scope with no major weakness; 9.5–9.9 almost no meaningful presentation improvement left. Do not use these preliminary scores to assert a market rank. A 10 requires no meaningful known improvement and complete evidence for the scope. Use one decimal per dimension. Overall = sum(score × weight)/100, rounded to two decimals; root verifies arithmetic.

Do not read the other judge's report or score. Do not infer what score the operator or root wants. Do not set a score to reward effort or to appease dissatisfaction. You are assessing the final rendered result. Cite exact frames for each dimension's deduction. Do not mistake a viewport edge for actual clipping, a deliberate partial carousel card for broken overflow, or an unchecked interaction for a proven defect. Identify uncertainty explicitly.


## Deliverable
Per brand per judge JSON under scores/<judge>/<slug>.json and concise report: dimensions keyed craft/identity/journey/trust/offer/mobile each score, rationale, evidence paths; overall, strengths, priorities, limitations, inspected_images. One decimal dimensions, two decimal weighted overall. Cite exact screenshot paths. If capture questionable flag root for recapture before penalizing the site. Do not read peer scores.