Dawn SA2 v3 (SOP 2.0) vs Alevia's longest-running natives — measured, 2026-09-03
Alevia side: the six longest-running long-form copies in the 8-12 TrendTrack pull (output/trendtrack-alevia-2026-08-12/ad-copies*.json, 99 copies, 91 over 2,000 words), sorted by days live: 116, 110, 109, 107, 103, 103. Two read in full (#1 senior-care worker, 116d; #3 nurse and her mother, 109d). Dawn side: v3 Fable and Sonnet from output/dawn-sa2-guilt-tof-2026-09-03/final/.
Measured
| Alevia #1 116d | Alevia #2 110d | Alevia #3 109d | Alevia #4 107d | Dawn v3 Fable | Dawn v3 Sonnet | |
|---|---|---|---|---|---|---|
| words | 3,159 | 2,593 | 4,668 | 4,410 | 4,610 | 3,998 |
| blocks | 112 paragraphs | 109 | 170 | 181 | 511 lines | 493 lines |
| words per block | 28 | 24 | 28 | 24 | 9 | 8 |
| avg sentence | 9.0 | 8.4 | 7.9 | 7.8 | 9.0 | 8.1 |
| brand reveal (of words) | 66% | 93% | 53% | 61% | 74% | 75% |
| dialogue markers | 64 | 44 | 58 | 92 | 90 | 121 |
| quoted direct speech ("...") | yes, throughout | yes | yes | yes | none (reported: "She said, ...") | none ("She said: ...") |
| timeline markers | 11 | 11 | 19 | 19 | 16 | 13 |
| prices named | 1 ($160/mo) | 3 | 1 ($80/mo) | 1 | 0 | 0 |
| numbers in body (years, readings, doses, minutes) | dozens | dozens | dozens | dozens | none by rule | one ("a Tuesday") |
| P.S. rungs | 0 | 0 | 3 + signature | 0 | 2 | 2 |
| emoji | 1 (👉 link) | 2 | 2 | 1 | 0 | 0 |
| analogies | 3 (old pipe, bucket with a hole, sandbags) | 4 | 3 | 3 | 1 (the full glass) | 2 (fire drill, match and kindling) |
| relationship cast | husband, daughter, granddaughter, mother, Helen | husband, daughter | mother, father, husband, daughter, Dr Anand | family + authority | son, Dana, one adult from a thread | son, Dana, two people from threads |
| reader-address questions | 20 | 13 | 15 | 27 | 5 | 11 |
| links | 1 | 2 | 3 | 1 | 2 | 2 |
Where the two match (the SOP moves Alevia also makes)
- The chain, then the reveal, then the results, then the guarantee. Both Alevia reads follow the SOP's order almost beat for beat: world → decline as small losses ("I stopped volunteering… I stopped traveling… I stopped cooking") → failed stack → inciting incident (the granddaughter and the swing; 153/97 on the cuff) → authority arrives → mechanism in dialogue with the narrator asking "and the beetroot?" "and the garlic?" → each alternative acknowledged, gapped, summarized ("Nobody fixed the wall.") → product named by the authority → day 3 / week 1 / week 3 / 3 months → the mirror of the incident (she pushes the swing) → guarantee → link. Our v3 does all of it.
- Reveal position. Theirs 53 to 93 percent, ours 74 to 75. Inside their band, on the SOP's target.
- Dialogue volume and reaction pacing. Their mechanism runs as a two-person scene with "I stopped walking." "My hands were shaking." between layers. Ours runs the same way ("I had to put the phone down for a second.").
- Small-company scarcity inside a sentence, refund in plain words, the mirror moment as the emotional peak. Same in both.
Where Alevia does something ours does not
- The narrator has an identity block, with numbers. "I'm 58. I've worked in senior care for 31 years. Lifting residents. Standing on concrete floors 10, sometimes 12 hours." / "I'm a nurse. 23 years." Both winners put age, job, and years of the problem in the first five paragraphs, before the scene. Our v3 narrators have no age, no job, no town, no years. We applied the no-qualifier rule to the narrator, where it only ever applied to the reader. Narrator numbers are texture, not qualification; every Alevia winner runs on them (12 years, 146/93, $160 a month, 4 months of beetroot, 2 PM crash, 22 minutes in the parking lot).
- A reader-address bridge right after the hook. Alevia #1 line 5: "I'm sharing this because… If you've been taking Lisinopril, Metoprolol, cutting your sodium, trying beetroot… this might be the most important thing you read this year." A list of the things the reader has tried, as the qualifier, then "here is who I am." Ours goes hook → scene with no bridge and no "I am telling you this because."
- Direct quoted speech. Every Alevia line of dialogue is in quotation marks and reads as a scene; ours is reported ("She said: …"), which reads as a transcript. Same content, colder register.
- A wider cast carrying the stakes. Husband eating dinner alone, daughter's "You're different," granddaughter's "Nana, come push me," father mowing over the tomato beds. Ours carries the stakes on the son and one adult from a thread. Our non-alienation rule removes husband and sibling; it does not remove a mother, a friend, a coworker, or the reader's own kid saying more than one line.
- The failed stack has doses, durations, and money. "Beetroot capsules for 4 months. Expensive pee." "$160 a month on supplements." Ours has the items and what happened, no durations, no money (our price ban, their price rule).
- Numbers in the results. 128/82, week 3; 126/77 at 11 weeks. Ours has no measurable by rule; "teeth on the first ask more mornings than not" is the closest.
- The close has more rungs. Alevia #1: promotion → small batches → 90 days → "But if it works?" conditional ladder (five "If by week 1…" lines) → "You'll wish you'd started years ago" → 👉 link → one coda line. Alevia #3: guarantee → link → scarcity → "You can keep doing what you've been doing… or" fork → link → "Thank you for reading this far" → signature → P.S. ×3 → link. Ours: loss line → guarantee → link → one emotional line → P.S. ×2 → link. Missing: the conditional future-pace ladder, the fork, the thank-you coda, the 👉 arrow, the signature.
- They say the honest thing out loud. "I won't tell you it was a miracle." "I'm not making promises. I'm telling you what happened." Our validator flags that class as provenance leakage; in Alevia it is the most-trusted line in the close.
- Three analogies, not one. Old pipe, bucket with a hole, sandbags. The authority owns the first two, the mother owns the third. Ours allows one.
- Paragraphs, not lines. All six longest runners are 24 to 28 words per paragraph. The SOP says one sentence per line; Alevia's live winners do not do that. Both are measured; the SOP is the ruling.
Where ours does something Alevia does not
- Ours never names a clinician; Alevia's authorities are a cardiac nurse and a cardiologist, and #3 signs "RN". Ours is the safer ban position; theirs is the higher-credibility one.
- Ours carries an owned object that returns (the dry toothbrush) and a P.S. that returns to it; Alevia #1 has no P.S. and returns the swing inside the results instead.
- Ours keeps the reader qualification to the card's categories; Alevia qualifies by a product list (Lisinopril, beetroot) that would be a P-item stack for us.
What to change in v4 (if you want it closer to Alevia without leaving the SOP)
- Add a narrator identity block after the hook, 4 to 8 lines, with her numbers (years of these mornings, her job, her age if you allow it). Reader stays unqualified; narrator gets specific.
- Add the reader-address bridge line after the hook ("I'm writing this because…" + the tried-it list in her words).
- Quote the dialogue. Quotation marks, scene register.
- One more character who notices (the reader's kid gets a second line; a friend or her own mother says one thing).
- Durations on the failed stack (how long she tried each), no prices unless you lift the ban.
- Close: add the conditional "if it works" ladder and the fork, the 👉 arrow before each link, one honest line ("I am not going to tell you it was a miracle"), and the thank-you coda.
- Allow a second analogy for the narrator herself.