## 1. Executive Summary

Arthur Pimentel Cordeiro ("Arthur PC") runs a direct-response nutraceutical operation selling into the US market from Brazil, and in this episode he walks through the machine behind reportedly 7-figure days (he claims passing R$1M in a single day at peaks). His formula has three legs: modelled (never invented) VSL copy built on the champion "secret lead" and validated mechanisms, "super-production" video currently powered mostly by AI avatars and licensed actor likenesses rather than live shoots and a creative testing engine running 25 30 new ads per day. Around those sits the unglamorous layer that actually protects profit: in-house support, fulfillment inspection, and account/BM infrastructure. Two lessons dominate the conversation. First, a great offer beats great creative: he prefers owning the offer over the ad, because a strong offer keeps converting when creatives die, while a weak offer cannot be saved. Second, everything is modelled and then improved copy from 3 5 winning competitor ads, mechanisms from validated proven ones with a renamed hook, stories from proven structures and then hardened through a revision pass that cut 15 minutes off a VSL and improved absolute results. His warning story is operational: a recommended fulfillment partner produced a 24% chargeback rate (~R$500k loss) from integration failures and undisclosed product instructions, which is why he now personally inspects deliveries daily and keeps refund decisions in-house.

## 2. Alpha / Sauce - what actually moves the needle, and exactly how to run it

- **Model, never invent.** Never write a copy 100% from your head; analyze at least 3 5 copies that are currently scaling in your niche (structure, lead, proof placement, offer block) before writing your first one. His first Brazilian copy was modelled from an offer in a completely different niche and still went to multiple 8 digits.

- **The champion lead is the secret lead.** State the secret up front ("the Egyptian glue "), build the story and mechanism, and only reveal what the secret actually is after the mechanism has persuaded. Most scaled copy today follows this pattern, usually dressed as "the X trick" (banana trick, cucumber trick, ice trick). One such copy he knows did ~R$70M.

- **Take a validated mechanism; rename the hook.** Do not innovate on the mechanism pick one that already scaled (even from another niche), rewrite it simpler, add demonstrative proof, and change the public-facing nickname/hook so you are not competing against the original. If you copy the running ad verbatim you are fighting the incumbent.

- **Demonstrative proof converts.** Build a filmed experiment that literally shows the mechanism: chitosan powder dropped into oil forming a ball that sinks (fat "grabbed and expelled"), detergent dispersing oil, a balloon demo. Even an illustration counts. If the viewer understands, they believe; if they believe, they buy.

- **Super-production = a face, done well.** A black background with a random voiceover does not sell; a credible, charismatic face does. He discarded the first actor mid-project for an unnatural "evil laugh" mid-VSL the test was: would the customer think this is garbage? Yes replace. The second actor, same copy, hit from the first take.

- **AI first, humans to validate.** He reached 7-digit days using AI video (HeyGen-class tools plus Regen-style avatars), buying actor likeness licenses around R$300 each for unlimited use, one actress charged R$6,000 and it paid off, and he has ~10 licensed faces. English AI voices are far more convincing than Portuguese ones. Real-actor shoots take too long; test fast with AI, upgrade to real people once a winner exists.

- **Offer block is the most under-tested lever.** One test moved average ticket from ~R$1,300 to R$1,700 and improved 6-pot conversion AND upsell take-rate. Almost nobody tests this. He now ships every new offer with the refined offer block format.

- **Offer > creative, but creative drives the P&L.** A good offer with mediocre creatives still sells; a good creative on a bad offer dies. That said, creatives control profit, consistency and scale: his first Brazil nutra run did R$2 3k/day at <10% margin; new creatives took the same offer to R$220k/day.

- **Diagnose creative bankruptcy fast.** If CPC hits R$30 60 or 50 creatives show no signal at all, stop insisting on the offer. Bad creative also hides whether the VSL is good you can't judge the VSL when the traffic quality is garbage.

- **Own support or eat chargebacks.** Unanswered tracking-code questions turn into same-day 7-day chargebacks. His rate went from 24% (disaster) to 2% today by answering complaints daily and refusing to let recovery agencies hand out refunds casually.

## 3. Methods & Setup - Step-by-Step

1. Pick the market first he sells only the US because Brazilian nutra is restricted by ANVISA recommendations and US buyers absorb $294/six bottles without pain; setup: choose a market with money and low regulatory friction.
2. Pick the niche by ease of hitting, not by dreams weight loss has huge audiences but CPCs of R$1 8; setup: audience size + urgent/strong pain (impulse-driven buy) + your budget for expensive CPC.
3. Spy 3 5 scaled copies in the niche (and pull swipe files from other niches, including 50-year-old classic swipes for angles nobody is using); setup: a swipe archive, not a blank page.
4. Choose a validated mechanism, check it has demonstrative-proof potential and is easy to explain to a layperson; setup: if it needs chemistry jargon, reject it.
5. Rename the hook/nickname of the mechanism and rewrite the explanation in your own words, simpler; setup: new curiosity, same proven core.
6. Build the lead: secret-lead format cite the secret, delay the reveal until after the mechanism; setup: story mechanism proof reveal offer.
7. Structure with the Stephan George framework as a skeleton (promise, mechanism, story, offer), but treat it as reference, not a checklist no real winning copy has all elements in order; setup: framework plus your swipe learnings.
8. Write the offer block from your best-performing version (the R$1,300 R$1,700 test format) rather than from scratch; setup: pre-baked AOV lift.
9. Produce the VSL: one charismatic actor (licensed AI face initially), documentary/interview setting, as many characters as the story needs for credibility; setup: quality bar = "would my customer think this is trash?"
10. Film demonstrative proof segments (lab-style or simple kitchen experiments) and cut them into the mechanism section; setup: proof before the price reveal.
11. Revise the copy: read it 1 3 times, cut boring/verbose stretches, compress 4 5 paragraphs into two sentences, add emotional wording and edit jolts for attention resets; setup: his optimized VSL lost ~15 minutes and made more money.
12. Launch creative production: 25 30 new creatives per day, mixing models, spy-derived material and your own regenerated formats; setup: creative volume is the throughput of the whole machine.
13. Watch CPC and signal: R$30 60 CPC or 50 flat creatives = stop; otherwise iterate hooks and leads; setup: pre-agreed kill criteria.
14. Stand up operations before scale: support person, daily tracking-code inspection, in-house refund control, BM/account structure with high limits and cash to fund 7-figure days; setup: contingency accounts grow limit by spending cleanly over days.

## 4. Tools & Stack - what each one is for here

| Tool | Role | In this episode |
|---|---|---|
| HeyGen / Regen-style avatar generators | Video production at speed and cost | Replaced live shoots; sold R$1M/day on AI faces; ~10 licensed actor likenesses at ~R$300 each |
| Licensed actor likeness deals | Premium AI production tier | Real actors sell usage rights for lip-sync; one actress charged R$6,000, still worth it |
| Swipe files (incl. 50-year-old classics) | Copy angle mining | Source of the Egyptian-glue secret lead and "trick of X" hooks |
| Spy tools + own avatar Instagram profiles | Market intelligence | Sees what the market actually sees, refreshes his own profile to keep spying angles live |
| Stephan George VSL framework | Copy skeleton | Structure base, optimized with internal winning patterns |
| Filimon's copy courses | Copy education | What finally made mechanism/story/lead concepts click vs. dry theory |
| Meta ads accounts / aged BMs | Scaling infrastructure | 7-figure days need high limits built by days of clean spending + cash reserves |
| Fulfillment partner (cautionary: Chip Offers) | US shipping | Integration failure 24% chargebacks, ~R$500k loss; 3/6-pot mixups 37% refunds |
| In-house support desk | Chargeback defense | Daily complaint chasing took chargebacks from 24% down to 2% |

## 5. Actionable Checklist

- [ ] Assemble 3 5 currently-scaling copies in the target niche (plus 1 2 from other niches) and map structure, lead, proof, offer block before writing anything.
- [ ] Write the secret-lead version: name the secret in line one, reveal it only after the mechanism.
- [ ] Pick a validated mechanism, rename its hook, rewrite the explanation simpler, add one demonstrative-proof video.
- [ ] Verify the mechanism can be shown on camera (experiment/illustration), not just claimed.
- [ ] Secure 1 3 AI faces (licensed likenesses) and shoot VSL in interview/documentary style.
- [ ] Kill any actor performance you wouldn't watch as a customer test with the "would I think this is garbage?" bar.
- [ ] Install your best-tested offer block (AOV R$1,300 R$1,700 format) into every new offer.
- [ ] Produce 25 30 new creatives per day; pre-define kill criteria (CPC R$30 60 or 50 creatives with zero signal abandon offer).
- [ ] Run the revision pass: cut 15 minutes of fat, compress dull passages, add attention-reset edits.
- [ ] Hire support before scaling and answer refund/traceability messages same-day; keep refund decisions in-house.
- [ ] Inspect fulfillment daily (tracking codes, hold lists, correct SKU/quantity shipped) his 24% chargeback disaster came from not looking.
- [ ] Build account/BM structure with contingency accounts and enough cash to fund a full 7-figure spending day.