What this is

A working pass at the improvements we discussed: every mock below is a faithful recreation of the live page with the changes applied — same copy, same design language, different decisions. Toggle “✨ Highlight improvements” on any mock to see exactly what changed. Everything is interactive; nothing is a static picture except the “current state” columns.

Where I'd start

Six workstreams and 63 findings is a lot — so here's the sequencing. Two bets first, both top-of-funnel acquisition plays: they're the cheapest friction cuts closest to the money, the fastest to put behind a test, and the clearest reads on new-customer CVR and CAC. Everything else below is here for depth.

One rule throughout: a signup lift that doesn't activate is a red flag, not a win — every test below carries an activation guardrail.

Six workstreams

• 1Floating mobile CTAs. No page on the site has a persistent mobile CTA — key pages run 12–20 screens deep with conversion out of view for most of the scroll. A primary + secondary floating bar is now in both page mocks. Homepage demo → Blog demo → Hypothesis: a persistent CTA converts intent that forms mid-scroll. Validate: A/B vs current, with a guardrail that hero-CTA clicks don't just shift. Metric: mobile new-customer CVR; secondary: bar CTR, checkout starts.
• 2Blog margin → product promotion. Article body is a 768px column pinned left; ~450px of desktop margin sits unused. Now a sticky promo rail + two mid-content CTA banners. Blog demo → Hypothesis: the empty rail can carry product promos without hurting reading. Validate: A/B rail on/off with read-completion as the guardrail; MVT banner placement. Metric: blog→product CTR, blog-attributed signups.
• 3Interactive experiences. They exist — a 5-step plan finder and a GLP-1 cost estimator — but both are hard to find, and the estimator asks for an email before showing the estimate. Two new ungated experiences built, plus the existing quiz surfaced prominently on the homepage. Interactive demos → Homepage demo → Hypothesis: value-first (ungated) tools complete more often and produce better-qualified leads. Validate: A/B gated vs ungated with the email ask moved after the estimate; watch lead quality, not just volume. Metric: estimator completion rate, Med+ signup CVR, CAC.
• 4Affordability on the homepage. Today, price appears once — “as low as $12/mo” in microcopy. Now a transparent pricing section with per-day framing, the 50% offer as a first-class card, and an insurance-check entry point. Homepage demo → Hypothesis: removing price uncertainty on the homepage lifts qualified plan-page traffic. Validate: section-level A/B; MVT the framing ($/day vs $/mo vs savings-led). Metric: homepage→plans CTR, new-customer CVR, 30-day activation as guardrail.
• 5Merged blog hero + title. Title block + standfirst + byline + date + hero image currently stack ~2 screens tall before content. Now one compact split header (~half the height). Blog demo → Hypothesis: a shorter header gets more readers into the content and to the first CTA sooner. Validate: template-level A/B. Metric: scroll-past-fold rate, read starts, first-CTA viewability.
• 6Full editorial scan. 63 findings across 9 pages in both viewports, each linked to the live page and capture — the depth behind the bets above. Editorial scan →

Where I'd focus first — highest-risk, highest-impact findings

A few things I noticed while digging into the site that I'd bring to a first working session — each framed as something to validate together and, if it holds, act on. All 63 findings, filterable by severity, page, and issue type →

• 1The headline stats and their footnotes don't line up — which in a medical category is worth a compliance look. I'd inventory every stat/footnote pair as step one. Detail: the homepage's “61% more body weight… at one month*” stat carries a single-asterisk footnote that describes a different analysis (a 92,160-member average-weight-loss read at 4 weeks); the only footnote that does describe Success Program participation — 166 survey responses from a 6-month trial — is marked ** and no stat on the page carries that marker. The homepage mock models the fix: every stat cites a unique symbol (*, †, ‡, §).
• 2The “free estimate in seconds” asks for an email first — email is the form's first field, with marketing-consent copy attached (verified live today). The most valuable friction cut available; I'd validate with an ungated A/B where the email ask comes after the estimate.
• 3Two lead offers run in parallel, not yet reconciled — “save 50% on a 12-month plan” (homepage, plans) vs “66% off your first month of Med+ / $25” (medication page). The Med+ pricing card does disclose “Then, $74/month”; the plans comparison table shows only “Starting at $25/first month^” with the ongoing $74 relegated to the footnote below it. Worth aligning into one price story; I'd test transparency-forward framing.
• 4The blog is off-message for the new WW story in places — the blog home's curated rails still surface recipe roundups titled with retired program names (“27 Low-PersonalPoints Recipes…,” 2019 and “19 low-SmartPoints value recipes…,” 2017), and two adjacent sections rendered seven cards with blank images — a content refresh plus a QA pass on the rails.