## 1. Executive Summary

Denis Pereira started in digital in 2014, while studying civil engineering (he dropped out in the second year). His first media was not Facebook but Google: back then he shot his own two-minute video with an old phone selling an importation course as an affiliate, put R$10/day behind it on YouTube, and made a R$150 sale within about three days. That first ad-powered sale hit different than any later revenue milestone. He spent the next few years bouncing around affiliate offers, Facebook ads, partnerships admitting that "looking sideways" at other people's results cost him years of compound growth. Around 2018 he left the "ownership of offers" path behind and started selling traffic management. His deepest learning phase was ~four years running the whole media buying for Dr. Rocha's operation, where he spent roughly R$150,000 300,000 per month on YouTube (around R$3 4 million total across the period) and learned YouTube at real volume.

Today Denis runs no offer of his own. He only does YouTube traffic for partners on a revenue-share basis (typically 10 20% of generated revenue) and claims this pays better than running his own product a large, still-saturated gap in the market for skilled direct-response YouTube media buyers. His core thesis: YouTube is a different media from Facebook; creative is roughly 80% of the result; the creative should run 1.5 2.5 minutes with the first five seconds decisive; Google accounts get "smarter" with time and a new account performs badly for the first 10 15 days no matter how good the offer; and the winning scale motion is validating one creative at a time on a tight budget before scaling 20 30% per step. He also details a second, riskier "aggressive" intraday scaling method that he only recommends to operations with strong cash reserves.

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

- **Creative is the 80/20 of YouTube, hands down.** Same offer, same funnel, something as small as a new first-five-second hook or a swapped visual has taken creatives from dead to profitable. His multi-seven-figure offers were almost always carried by one creative that broke R$100 200k+ in revenue; one operation even poured ~R$3 million into a single winner.
- **Hook length is a myth until it is not:** most scaled creatives ran ~2 minutes (sweet band 2:00 2:30, floor 1:30). The five seconds at the front are everything he has kept the narration identical and changed only the visual of the first five seconds and revived a failing creative.
- **Congruence with the VSL is non-negotiable.** Pull the lead/hook from the VSL, mirror it in the creative, and VSL retention climbs. Point people at "a banana" and the landing page must absolutely deliver a banana lack of congruence is his biggest recurring complaint about the market.
- **Work the "media translation"**: Facebook-style short home-shot creatives (even a terrible 10-second one can win there) fail on YouTube ~99% of the time. YouTube needs longer, deeper, more produced copy; the reverse direction (YouTube creative onto Facebook) sometimes works, so shoot for YouTube first.
- **Open the mechanism loop, don't explain it.** The best creatives carry at least a brushstroke of the product's mechanism a provocation that opens a curiosity loop without revealing the mechanism itself.
- **New Google accounts are meant to be bad.** The account/pixel accumulates the intelligence; the first 10 15 days are rough regardless of offer quality, which is why 99% of people quit. A seasoned account in one niche beats Facebook-level delivery and validates almost anything you throw at it he saw campaigns cross ~3,000 in-account conversions where "villainous" volume made results automatic.
- **Do more, do better, then do new** the Hormozi framework he applies religiously. Find something that converts, then scale that same motion laterally with more creatives before inventing a new funnel. Switching lanes early kills the compound effect; he says his own 9-year detour is why he is not 10x bigger.
- **Contrast with the comparison trap** revenue is the market's favorite show, but nobody shows profit. "The shortcut is doing the right path and not deviating."

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

1. Accept that Google and Facebook are different media before opening Ads Manager; the first question every mentee asks is "can I reuse my Facebook creatives?" and the honest answer is "yes, but don't" build YouTube-native creative instead; setup: run this conversation as the first tutoring session, no account touched yet.
2. Write deeper copy than Facebook demands a developed argument, not a 30-second skit; minimum realistic length is 1.5 minutes and most winners live at 2:00 2:30; setup: draft the creative script with a persuasion arc before producing anything.
3. Write the first five seconds as its own micro-asset; treat it as the "lead" of the creative and expect to iterate it in isolation (keep copy, swap visual: recipes, expert on camera, AI avatar); setup: produce each creative as a base video plus 5 10 alternate cold-open variations.
4. Mirror the VSL's opening in the creative so the click-through feels seamless; reviewing the VSL lead to mine hooks is his default production step because tested congruence raises VSL retention; setup: transcribe the VSL lead and align the creative promise/mechanism teaser to it.
5. Include a mechanism provocation name the mechanism's existence without revealing it to open a loop that only the VSL closes; setup: add one "mechanism hint" beat per creative.
6. Build a batch of creatives before validating; a large operation he runs tests 200 300 ads in a month, currently amplified by the stricter policies; setup: set a weekly creative-production target proportional to budget.
7. Run the delivery test: launch the creative and check that spend lands where you want it, not leaking into the display/partner network, which destroys intent and is a classic new-policy failure state; setup: a small spend window to inspect placements before committing budget.
8. Validate each creative on a "one ticket over five days" budget if the product sells for R$200, give it R$200 over 5 days and watch for any sales signal; setup: fixed R$=1 ticket budget, 5-day window, no interference.
9. If it shows a pulse, make variations of that creative and run them through the delivery test again; setup: variant batch from the promising concept, re-tested the same way.
10. Scale the validated winner 20 30% per step until ~R$1,000/day, then keep scaling 20 30% while the CPA holds; pause and step back down to the last profitable stage the moment Google flags elevated spend; setup: laddered budget increments with a stop rule tied to CPA.
11. Reserve the aggressive intraday scale big launches, doubling budgets (e.g., R$3,000 opening a day climbing fast) only for operations with cash, a known offer rhythm, and an already-smart account; never run it on multiple live campaigns at once; setup: separate experimental campaigns, never touching profitable ones.
12. Stay current through a peer circle or mastermind; real knowledge is "what works right now," and when Facebook stabbed everyone in February-style updates, the groups who kept operating (like the masterminds he talks about) recovered fastest because some people never paused; setup: join/maintain a small group of daily operators in the same media and compare notes weekly.

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

| Tool | Role in this operation |
|---|---|
| Google Ads (YouTube) | The core buying platform; he manages whole accounts and campaigns for partners, spends R$150k 500k/month on it at scale |
| Facebook Ads | The comparison baseline he left; useful for validation and as an extra conversion channel, but its creatives/playbooks do not translate to YouTube |
| Account/Pixel historical data | The real "smart tool" Google optimizes per account; seasoned accounts validate almost anything, new ones burn budget for 10 15 days |
| VSL landing page | The conversion close; creative must mirror its lead for congruence and it holds the "mechanism" reveal that the creative only teases |
| Synthesia / HeyGen (AI avatars) | Lets traffic-only operators run talking-face creatives without a filmed expert; excellent in English and Spanish, still slightly stiff in Portuguese |
| Mastermind / operator community | His intelligence source for "what works right now" after policy changes faster iteration than solo testing |
| Redirect/test-of-delivery check | Verifies placement quality and flags display-network leakage before a creative earns real budget |

## 5. Actionable Checklist

- [ ] Write YouTube-native creative copy (1.5 2.5 min), never paste a 30 40s Facebook creative.
- [ ] Draft 5 10 alternative five-second cold opens per creative before production.
- [ ] Mirror the VSL lead/hook in the creative for congruence.
- [ ] Include one mechanism-tease beat (open the loop, don't close it).
- [ ] Produce at least a small batch (e.g., 5 10) before first validation big operations test 200 300 ads/month.
- [ ] Run a delivery test and confirm spend is not leaking to display/partner placements.
- [ ] Validate creative on "one ticket of budget over five days."
- [ ] On any sales signal, clone variations and re-run the delivery test.
- [ ] Scale +20 30% per step to R$1,000/day, then continue while the CPA holds.
- [ ] Pause and return to the last-profitable budget if Google flags massive spend.
- [ ] Leave the aggressive intraday scale to cash-rich operations on already-optimized accounts isolate it in separate campaigns.
- [ ] Track claims, not lifestyle: measure profit, not shown revenue, and ignore what other operators "look" like they make.

## Post notes

- Interview: Segredos da Escala #017, host Joao Campos, guest Denis Pereira.
- Denis describes lifetime spend of ~R$3 4M on YouTube across a ~4-year period with Dr. Rocha, then continuing as an external media buyer; figures are self-reported and unverified.
- The claim that AI review cannot read VSL audio and that this creates an approval "gap" is his hypothesis, not established fact.