## 1. Executive Summary

Wellington Costa (founder) and Lucas Canavarro (operations director) run a direct-response operation that has grossed over R$50 million in digital and reached R$1 million per day selling into the United States. Their differentiator is not a niche or a creative genius - it is automation. They built an in-house traffic automation layer that launches up to200 campaigns in under three minutes, pulls every Facebook metric into Google Sheets through APIs, and replaces a team of three to five junior traffic managers with a single manager who reviews charts instead of typing spreadsheets. The episode walks through the full stack: profit-based budget rules that double winning campaigns from R$50k to R$500k in a day, a stop-loss protocol for sudden spend spikes, the reversal of the "reset budgets at midnight" habit, the HR system (PDI plus the three Cs) that keeps a lean 17-person team executing, and the hiring thesis behind it all - find a person with your own background and clone yourself. It also covers the conflict-of-interest question of running a SaaS tool while still running the operation that feeds it, with data encrypted so the operation cannot see other operations' information.

## 2. Alpha / Sauce

- **One manager replaces three to five.** Before automation, scaling meant hiring more junior managers to upload campaigns by hand. Their system mass-launches up to 200 campaigns in less than 3 minutes, so a single manager now runs what used to take a full desk of people. Hiring stopped being the scaling lever; code became the lever.
- **Automate the spreadsheet, not the judgment.** The traffic manager used to fill in daily spend per ad account by hand. A Python script now pulls Facebook data through the API straight into Google Sheets - no typing, no transcription errors. The manager's freed hours go to strategy: which creative to push next.
- **One chart decides budget moves, hourly.** A RedTrack-fed sheet renders an hour-by-hour profit chart per campaign. The rule: campaign sold in the last two hours and profitable - raise the budget; bleeding - cut it in half or to 30 percent. They check every one to two hours, not once a day.
- **Scale budgets geometrically, not linearly.** A campaign starting the morning at R$50,000 goes 50k, 100k, 200k, and has closed the day at R$400,000-500,000 in a single campaign budget. Starting big at the auction ("rolling up sleeves at the auction") means the campaign already has R$100k in the bag by mid-morning.
- **Reverse the midnight reset.** The market habit is to slash budgets at midnight and start over. When their chart showed profit climbing overnight, they deliberately left budgets alone - and profit kept compounding. The stop-loss: when a spike burns too much (roughly R$40k in an hour), cut the budget to 20 percent of peak, then restore to normal levels once profit recovers. Most of the time this worked.
- **Creative-level metrics in one sheet.** Another automation pulls hook rate, body rate and CTR per creative into a shared spreadsheet - the same view they demo at events - so creative decisions come from one screen instead of five dashboards.
- **CPM advantage from volume.** Their scale lets them buy infrastructure and server-side capacity at a price competitors cannot match, and they front-ran Brazil's tax reform: a 12 percent tax on ad spend (R$100k spend implies R$12k) was anticipated eight months out, so the operation was restructured before it bit.
- **Clone yourself when you find talent.** Wellington's hiring rule, borrowed from a TV show about a billionaire proving he can rebuild from zero: when you find a real talent, make room for him. He hired Lucas as a project manager with an explicit brief - "your spot is to become my clone" - because Lucas had the same background (engineer, finance, analytical, executes fast). Lucas dissected the Vurb platform on day one and implemented it on day two; he finished a R$20,000 ClickUp mentorship in a week and knew more than the instructor.

## 3. Methods and Setup

1. **Map the manual work first.** List every recurring task a manager touches daily; setup: campaign uploads, spend spreadsheets, creative reporting, billing contingency.
2. **Kill data entry with API pulls.** Automate anything that is transcription; setup: Python script hitting the Facebook API writing directly into Google Sheets, plus a RedTrack API feed for conversions.
3. **Build the hour-by-hour profit chart.** One graph per campaign showing spend, revenue and profit for the current day; setup: refreshed on demand so a decision can be made in the two-hour window.
4. **Write explicit budget rules.** Profitable in the last two hours: raise. Overspending vs checkout tolerance: cut in half or by 30 percent; setup: written rules so the manager executes instead of debating.
5. **Start campaigns with real budget.** Launch winners at R$50k/day scale rather than trickling R$5k; setup: auction momentum by mid-morning, then double: 50k, 100k, 200k.
6. **Install a stop-loss at 20 percent.** On a spike loss (about R$40k burned in an hour), drop the budget to a fifth of peak; setup: automatic alert threshold on the chart, restore only when profit trend is back.
7. **Overrule market habits with data.** If the chart says profit rises overnight, do not reset budgets at midnight; setup: a standing rule that beats the convention of "start the day clean."
8. **Consolidate creative metrics.** Hook rate, body rate and CTR per creative in one sheet; setup: automated pull, shared access, reviewed in the creative meeting.
9. **Mass-launch through the API.** Replace hand uploads with scripted campaign creation; setup: templates that replicate a validated campaign structure in bulk - up to 200 in under three minutes.
10. **Centralize contingency.** A second spreadsheet pulls every bill and ad account status through the API - available slots, account health, overview; setup: contingency plan visible before an account is lost, not after.
11. **Hire the clone, not the resume.** Look for someone with your own profile (analytical, technical, from your field) rather than the polished mid-level candidate from a big company; setup: the "mid-level" profile interviews well and executes poorly because they never lived through an exponential-growth moment.
12. **Run a one-week trial before the contract.** Culture and execution show up fast when the person is actually doing the work; setup: plus logic, CAS and skills tests for the theory filter.
13. **Give everyone a PDI.** Individual development plan every six months, written in a system only the employee's login opens, structured with the three Cs - what to stop doing, keep doing, start doing; setup: promotion or cut is decided against that document, not against memory.
14. **Pay for the courses they ask for.** When someone brings a course they want, the company buys it (a R$20k YouTube mentorship, a copy course); setup: the ask itself is a signal - never asking is a red flag, as is repeating the same mistake twice.

## 4. Tools and Setups

| Tool | Role | In this episode |
| --- | --- | --- |
| Facebook Marketing API | Campaign mass-launch + data pulls | 200 campaigns in under 3 minutes; fills Sheets automatically |
| Python scripts | Glue between API and Sheets | Replaces manual daily spend spreadsheets |
| RedTrack | Conversion tracking API | Feeds the hour-by-hour profit chart |
| N8N | Automation bus | Where part of the team routes API jobs |
| Google Sheets | Decision surface | Profit chart, creative hook/body/CTR sheet, billing contingency |
| ClickUp | Project management | Lucas ran ops on it after a one-week mentorship |
| Vurb | VSL player | Revenue and profit tripled the day it was installed |
| PDI system | People development | Six-month plans, three Cs, employee self-service |

## 5. Actionable Checklist

- [ ] Every recurring manual task listed and classified as automate or keep
- [ ] Facebook data flowing into Sheets with zero typing
- [ ] Hourly profit chart live for every campaign
- [ ] Budget rules written: raise on profit, cut on overspend
- [ ] Launch budget for winners set at 50k, not 5k
- [ ] Stop-loss set at 20 percent of peak budget
- [ ] Midnight reset habit overruled by the profit chart
- [ ] Hook rate, body rate and CTR consolidated in one sheet
- [ ] Campaign templates ready for bulk API launch
- [ ] Ad account contingency pulled via API, not memory
- [ ] One-week trial used before any full-time contract
- [ ] PDI with the three Cs active for every operator