Healthtech A/B tests we've caught

The hero headline differs across visits: visits 1 and 2 (Safari/Chrome) show 'Virtual therapy that works', visit 3 (Firefox) and visit 5 (Chrome) show 'Research-backed therapy, covered by insurance', and visit 4 (Edge) shows 'Space to figure things out'. Since multiple different browser engines (Chrome, Firefox, Edge) show different headlines, this is consistent with a real A/B test keyed on user/session rather than browser engine.

Visit 2 (Chrome) was routed to a different funnel step — the 'insurance-not-eligible' dead-end page showing 'Sorry, government insurance is not eligible at this time' with a 'Got it' CTA — while all other visits (Safari, Firefox, Edge, and a second Chrome) were routed to the 'insurance-type' question step asking 'What type of insurance do you have?' with six answer options. This is a confirmed server-side funnel split backed by divergent final URLs and differing experiment bucket values in localStorage.

Visits 1 and 2 (Safari and Chrome control bucket) show a minimal image-light layout with large blank/dark placeholder sections and no lifestyle photography, while visits 3, 4, and 5 (Firefox, Edge, and Chrome variant bucket) display a rich image variant with a '20%' stat hero section, real patient lifestyle photos, a doctor image, and additional social proof imagery throughout the page.

Visit 1 (Safari) shows different copy variants across multiple sections compared to all other visits: the hero subhead reads differently, the 'Lowering costs' section headline differs, the 'Making healthcare' section headline differs, and the testimonials section headline differs. Visits 2–5 (Chrome, Firefox, Edge, Chrome) all show consistent copy, while Safari alone shows the control/alternate variant.

Visits 1, 2, 3, and 5 show an insurance-type selection step ('What type of insurance do you have?') with six answer choices, while visit 4 (Edge/server-side routed to a different URL) shows a downstream cash-pay options page ('We have options that don't require insurance') with a single Continue CTA — confirming the server-side funnel split described in the context.

Visits 2, 3, and 4 (Chrome, Firefox, Edge) show a significantly richer page with additional sections: a '14%' weight loss stat block, a 'Meet the Wegovy pill' hero section, a 'Wegovy two ways' comparison table (pen vs pill), lifestyle/member photos in testimonials, and a doctor image in the membership section. Visits 1 (Safari) and 5 (Chrome) show a stripped-down version missing all these sections. Since visits 1 and 5 share the same minimal layout despite being different browsers, and visits 2/3/4 share the richer layout across different browser engines, this is a real A/B test — not a browser rendering difference.

Server-side routing splits visitors between two distinct funnels: a minimal variant (/n/) with no pricing callout banner and stripped-down content, and a richer variant (/p/) featuring a 'from $199/mo' hero price banner, a '19% average weight loss' social proof stat, a 'Meet Wegovy HD' product section, a side-by-side 'Wegovy two ways' pen vs. eHD comparison table, and testimonial photos with real member images.

Visit 2 (Chrome) shows additional section headlines not present in visits 1 (Safari) and 3 (Firefox): a large hero subheadline 'Improve your health — now and for the future' with supporting copy, a 'Compare the lineup' section header, an 'Experience Peak membership' section header, a 'What's included' section header, and a 'Start free, choose later' section header. Visits 1 and 3 share the same layout without these section headers.

Visit 2 (Chrome) shows a section header 'Start clearing acne in 3 steps' above the 3-step process icons, while visits 1 (Safari) and 3 (Firefox) do not display this heading. Since Safari and Firefox both lack this heading and only Chrome shows it, this appears to be a real content difference rather than a browser rendering artifact.

The hero headline differs: Safari and Chrome show 'The wearable designed for lasting progress' while Firefox shows 'The wearable built to improve you.' Divergent experiment UUIDs in cookies confirm different bucket assignments.