CHL page vs Seed DS-01, blind judged

Verdict: close. The judge's own overall score: Seed 8.0, our default page 7.1, our fog page 7.7, the Seed-structured mock 8.0. Seed edges the default page, and is level with the fog page and the mock. Judge noise is about 0.5.

Seed leads on look (8.7 vs 8.2), trust (7.8 vs 6.7) and reviews. We lead on conversion (7.8 vs 7.0) and ease on a phone (7.7 vs 6.8).

The one section Seed clearly wins is reviews (3 of 3 judges). It shows a bar for every star count, search and topic chips. We cannot copy it without the real star counts and reviews. The mock added five Seed-style blocks from facts we already have and scored 8.0 in all three runs (our default page 7.3 in that round). That is three runs, suggestive only.

Newest, r51 = r50 with a new first question in the FAQ: "Can I take it with my antidepressant?" Yes. Everyone in the study was already on an antidepressant, and they took 15 mg of L-methylfolate alongside it every day for 60 days. (The blind judges said this was the most important unanswered question for this shopper.) Open r51. Nothing is on the store.

Newest, r50 = r49 with the long desktop headlines fixed: on a computer screen each headline stays in its own column but now fits in three lines or fewer (it was up to six), and those sections split evenly so the text side has more room. Phones unchanged. Open r50. Nothing is on the store.

Newest, r49 = r48 with a desktop pass and nicer review cards. On a computer screen, every section below the buy box now uses the full width in two columns (headline left, table, study, timeline, chart, tests or reviews right), like Seed; the three detail cards sit side by side; the reviews keep their summary, search and topics on the left while you scroll the reviews on the right. Review cards now have initials and a topic tag. Phones are unchanged apart from the review cards. Blind desktop judges: new vs old won 3 to 2 with every score up (looks 6.6 to 7.6); new vs Seed's desktop page won 5 to 0. Open r49. Nothing is on the store.

Newest, r48 = r47 with your 100 customer reviews in place of the old review cards: 4.9 from 2,600+ reviews, a breakdown of the 100 (86 five-star, 13 four-star, 1 three-star), search, topic buttons with counts, three reviews on phones, and the rest loading only when someone scrolls near or taps. The page itself got slightly smaller. Open r48. Nothing is on the store.

Newest, r47 = r46 with the review count reworded everywhere to "4.9 from 2,600+ reviews" (a blind judge said "2,600 five-star reviews" sounded inflated; you agreed). Open r47. Nothing is on the store.

Newest, r46 = r45 with every line of copy checked for AI tells. Four fresh blind readers (Claude Sonnet 4.6 twice, Claude Sonnet 5.5, GPT-6 Astra) read the whole page next to Seed's copy and listed every line a real person would not write. Flagged lines were rewritten as plain sentences, using real customer wording from L-methylfolate reviews and Reddit; every fact, price and term is unchanged. Before: average 5.5/10, 6 lines flagged by all four readers, 11 by three. After: 6.75/10, none by all four, one by three (fixed). Seed's copy scored 2 to 4. Kept on purpose: your hero and free-bottle headlines. Open r46. Nothing is on the store.

Newest, r45 = r43 with the closing section redone (you said everything else is good). Headline "You’ve felt halfway / long enough." in the page’s headline style, picked by three blind Claude Sonnet 4.6 reads out of five options (the old "Care for the brain you count on." came last in all three). Deeper green, side-by-side plan choices, FREE and the old price on one line, the reviews line shown, a shield on the guarantee. Five blind judges compared old and new closing: new won 5 of 5 (conversion 8.0 vs 6.0, looks 8.0 vs 6.6, copy 8.0 vs 4.0). Open r45. Nothing is on the store.

Newest, r43 and r44. Both have graphs that start as soon as the card is on screen (done in about half a second) and swipes that start the next graph right away. The headline now also fits on phones set to reduce motion (it ran off the screen there before).

r43 keeps the page order you have now, with the free-bottle section after the 60-day timeline. Open r43.

r44 is the order five fresh blind judges built when asked to reorder the whole page (every section kept): top and buy box, details, the study (32% vs 15%), meds need active folate + table, full label + lab report + guarantee, reviews, two ingredients, why it's free, creator videos, every lot tested, 60-day timeline, FAQ, one daily habit, community videos, closing offer. Their shared logic: proof a skeptic can check before testimonials. Open r44.

Then five more blind judges compared r43 and r44 head to head: r44 won 3 to 2, every judge called it close, and both got the same conversion score (7.0). So the order is not what will move sales. Nothing is on the store.

Headline fit (all four landers, before and after) · Seed's block vs ours, side by side · Open mock r29 (Seed-style) · r27 (options A and B) · r25 · r24 (scroll-driven timeline) · r23. Nothing is live on the store.