CHL Wonder: fresh design research and build brief

Research snapshot: 2026-09-12 UTC
Decision status: The fresh concept is complete. Final browser QA passed 55/55 checks, and the bounded visual review disposition is ship after all five review findings were resolved.
Review previews: CHL Wonder editorial concept r1 · Research report. These are immutable review previews, not a Shopify or production release.

Executive finding

The current v61 page is not short of information. Its problem is that nearly every kind of information receives similar visual treatment: persistent pale green/tinted panels, repeated rounded containers, one Arial-like voice, modest differences between large and small type, and a dense purchase box. That flattens the sales journey. Product, mechanism, reassurance, social material, and offer begin to feel interchangeable even when the content is useful. The warm CHL photographs carry more personality than the surrounding layout.

This diagnosis is an expert design inference from the captured v61 desktop and mobile states, not a measured conversion result. Nothing in this research establishes that a reference brand is profitable, that its page converts well, or that adopting a pattern will improve CHL performance. The purpose is to form a testable new design hypothesis.

The new hypothesis is a warm, editorial supplement page that feels specific to CHL. It keeps emerald and sage, but does not merely swap the existing template for the familiar cream-background/serif template. The opening uses the exact pain-first line “For the days you lose your train of thought.” at meaningful scale, paired with a natural-light bottle photograph. Product identity, 15 mg strength, and the exact offer remain easy to find early. Asymmetric composition, deliberate shifts in density, Fraunces Roman for expressive display copy, and Manrope for functional text create hierarchy. Capsule macro photography, routine context, and the literal Supplement Facts panel provide material proof.

What was observed and what is inferred

The trust rationale is also a hypothesis grounded in established UX guidance. Nielsen Norman Group describes four contributors to perceived trustworthiness: design quality, upfront disclosure, current and comprehensive content, and connection to the rest of the web. That supports a polished page with visible terms, complete product facts, and real source links; it does not prove CHL conversion lift. Nielsen Norman Group: Trustworthy Design

Baymard reports that shoppers use product images to explore and understand products, and that short descriptive captions can clarify what each image demonstrates. For CHL, the practical inference is to caption the capsule-in-hand view for scale, the macro for form, and the facts panel for inspection. Baymard: Product Images Need Descriptive Text Baymard's broader PDP research also emphasizes sufficiently large product images and documents findings from ecommerce usability testing; the candidate should treat the gallery as decision evidence rather than decoration. Baymard: Current State of Ecommerce Product Page UX

Fourteen references, with evidence limits

The evidence records seven successful browser visual sets: Methyl-Life, Seeking Health, Needed, OSEA, Graza, Ritual, and Dieux. Seed was stopped by a security verification page; MethylPro returned “Store unavailable.” The remaining five references below are based on page-visible or search-indexed text captured in the notes, not a full visual audit. Ritual and Dieux were successful replacement captures recorded in notes.md and capture-replacements.mjs; the original seven-site sources.json manifest remains preserved rather than rewritten after the fact.

Candidate design contract

The candidate should open with the pain statement, product name and 15 mg identity, then let a natural-light bottle image carry the first impression. The free-first-bottle offer is prominent but is support for the shopper's need rather than the sole premise. Preserve the exact commercial terms: subscription is $0 plus $9.99 shipping today, then $50.99 every two months, with skip, pause or cancel; one-time purchase is $59.99 plus $9 shipping, $68.99 today, with no recurring orders; preserve the 60-day money-back guarantee and the existing sold-out/fail-closed behavior.

The page must also retain the existing high-dose guidance and professional-advice language, including pregnancy/breastfeeding and medicine/supplement considerations, and must not present this as a complete prenatal. Product-specific claims and public-health source links stay bounded to evidence already approved for CHL.

Use Fraunces Roman sparingly for large pain, section, and editorial moments; use Manrope for product data, controls, labels, terms and body text. Emerald and sage remain primary, with warm neutrals and natural photo color. Vary section density and alignment, including open white space, an asymmetric image/copy composition, a compact proof band, and a materially distinct offer surface. Avoid a full clinical-white system, neon playfulness, SaaS feature-card grids, and motion or gradient decoration used merely to signal polish.

Trust material requires an evidence policy. v61's release manifest says it preserves twelve reviews, while the review forensic summary found no verified imported corpus and classified available provider records as test or sample/unverified. Therefore those twelve quote cards are internal/sample content, not authenticated customer evidence. The new candidate must omit their names, dates and quotes. The “2,600 five-star reviews” count is operator-provided and has not been independently audited; label it accordingly in internal receipts and do not strengthen the claim. Retain only the five selected creator/UGC modules with verified source/product match, showing the provided-product or affiliate disclosure wherever applicable. Build metadata, evidence labels and QA language belong in receipts, never in the shopper-facing UI.

Skill research and actual use

Skills can improve process consistency; they are not evidence that the page will convert better.

Full paths, commit hashes, local compatibility changes and smoke-test results are in the local artifact skills/INSTALL-RECEIPT.md. The UUPM runtime/data suite passed 133 tests; its search, data validation, local scripts and dependency resolution were smoke-tested. The receipt also records four broader upstream maintenance-test import errors caused by the absent full-repository layout; they do not exercise the distributed runtime and remain disclosed.

Measurement plan

Before any separately authorized production release, compare control v61 with the new concept using concurrent randomized allocation and the same eligible traffic, offer and media. The primary outcome is completed purchases divided by eligible landing-page visitors. Secondary diagnostics are checkout starts, offer comprehension, and creator-video use. Monitor refund and cancellation rates plus page performance, since a higher free-offer take rate is not evidence of sustained conversion if later cancellations or refunds rise.

Do not assign an uplift target, aesthetic score, power level or sample size in advance without baseline data. Derive the analysis plan and required sample from observed control conversion and business constraints, then hold allocation and definitions stable. This is a proposed experiment; no test has been deployed.

Artifact and QA ledger

Evidence was collected without purchase or form submission. Live storefronts, indexed text and availability can change. Screenshots record a moment, blocked captures do not support visual conclusions, text-level references do not support layout judgments, and competitor claims are observations of what the pages state rather than independent validation. No competitor supplement claim should be transferred to CHL.