## 1. What happened

Anthony Carreiro spent years making creatives for other operators, then went affiliate on his own. He lost for five months on Meta, switched to Taboola, and says he hit R$400,000 in a single day there with no team. The episode is the operating detail behind that switch: why he left Meta, the only two variables Taboola gives you, the pre-sell structure he validated in more than ten niches, the spy routine, and the cash and team rules he uses to keep the operation small.

## 2. Alpha

- Traffic source is the first decision, not the creative. In his read of 2026, Meta is the source giving the most problems; Taboola native is the one still stable enough to scale. [source switch]
- Taboola gives you two variables: the headline and the image. He keeps the image and changes the headline across weight-loss, rejuvenation, memory and ED, so one creative asset can be tested across niches. [2 variables]
- The pre-sell is a qualification layer, not a short bridge. Taboola clicks are cheap, so a longer pre-sell that qualifies the lead is worth more than a short one that buys the same cheap click. [long pre-sell]
- His framework has a fixed order: hook (real image plus headline), qualification (interview-style pre-sell), mechanism of the problem, then the solution. He says the same format worked in ten different niches. [10 niches]
- Split tests into micro and macro. Micro tests the creative and the offer; macro reads the real numbers after launch: conversion, CTR, CPC, CPA. Do not judge a creative before the macro numbers exist. [micro/macro]
- Spy every day. The success metric is duration: an ad that has been running for a year beats one that started an hour ago, and a million copies beats one copy. The date range and the paused status tell you what is still working. [duration]
- Scale on Max Conversions with a CPA target. Pass the Taboola pixel so the system can optimise for CPA; if you leave it free, the system picks its own target. [CPA]
- Cash is a pillar, not a detail. With network and knowledge already in place, around R$20,000 is enough to test the first campaigns. Platform withdrawals three times a week let you start with less than a full media budget. [R$20k]
- Three pillars before going affiliate: cash box, networking, knowledge. If one is missing, the other two have to carry it. [3 pillars]
- Keep the team at zero until the operation is proven. He started alone, went to two, tried ten, and came back to three. [team]

## 3. Method & setup

1. Pick the traffic source first. If Meta is the bottleneck, test Taboola native before rewriting the offer.
2. Build the pre-sell in the order that worked: hook (real image + headline), qualification (interview), mechanism, solution.
3. Write the pre-sell long enough to qualify. Cheap clicks are not the goal; qualified clicks are.
4. Hold the image constant and test the headline across niches. Then test a new image against the winning headline.
5. Before launch, spy the network daily. Filter by run duration, copies and date range; model only ads that have survived for months.
6. Launch with Max Conversions and a CPA target. Install the Taboola pixel if you want the platform to optimise to CPA.
7. Read micro results first, then macro. CTR and CPC are intermediate; conversion and CPA decide whether the campaign lives.
8. Fund the first tests with roughly R$20,000 if network and knowledge are already in place. Use the platform's payout rhythm to recycle cash.
9. Keep the team at zero until the numbers are repeatable. Add people only after the operation proves it can carry them.

## 4. Tools table

| Tool | Use |
| Taboola | Native traffic source; the channel the whole method runs on |
| Taboola pixel | CPA targeting and campaign optimisation |
| Taboola spy tool (name garbled in the transcript) | Daily creative research: duration, copies, date range, paused status |
| VTurb | Video host and player; the podcast's sponsor, mentioned in passing |
| Meta Ads | The source he left after five months of losses; used as the contrast case |
| YouTube | Where the episode and the interview live |

## 5. Numbers table

| Metric | Value |
| Best day | R$400,000 in one day, operator-stated, no team |
| Time lost on Meta | 5 months before the switch |
| Time to first result on Taboola | First month |
| Niches tested | 10+ |
| Starting capital | ~R$20,000 with network and knowledge in place |
| Withdrawals | 3x per week on the platform |
| Team | 1 -> 2 -> 10 -> 3 |
| Taboola variables | Headline and image only |

## 6. Checklist

- [ ] Traffic source chosen before the offer is rewritten
- [ ] Pre-sell built in the order: hook, qualification, mechanism, solution
- [ ] Pre-sell long enough to qualify, not a short bridge
- [ ] Image held constant while headlines are tested across niches
- [ ] Spy routine run daily, filtered by duration and copies
- [ ] Max Conversions plus a CPA target set at launch
- [ ] Taboola pixel installed if CPA optimisation is the goal
- [ ] Micro tests read before macro numbers
- [ ] ~R$20,000 test budget in place, or network and knowledge to start with less
- [ ] Team kept at zero until the numbers repeat

## 7. Proof

- Video: https://youtube.com/watch?v=3h6YsOiXvIk
- Source file: `VIDEO - O Afiliado Que Escalou 400K Dia Sem Equipe Anthony Carreiro Segredos da Escala 170 [3h6YsOiXvIk].md`
- Reliability: low-medium. The source is an auto-translated Portuguese transcript with garbled tool names and numbers; every figure is operator-stated, not independently verified. The transcript renders Taboola as "tabula" and VTurb as "Veturb"; read those as the named platforms. The spy tool's name is too corrupted to repeat, so it is described by function only.