## 1. Executive Summary

Kaio Larroque is a native-ads operator on Taboola and Outbrain who, by his own account, has invested around R$15 million into native networks and never once launched a Facebook campaign. He came up through an internship at an office, discovered affiliate marketing during the pandemic, and got pulled into native by Mateus Rockenbach after the two crossed paths starting with Google Display Network nutra campaigns at a R$40 CPA and then moving to the native networks when display stopped working. His central argument is that native is the stable, boring, reliable source: Facebook and Google swing wildly and block everything, while native keeps producing for a good offer with a renewed funnel. He runs everything through a control-and-test workflow always keeping about 20% of traffic on new variables and beating the control and he and his team pushed roughly R$1.2 million per month through Taboola and about R$200,000 per month through Outbrain at peak. The R$15M, the monthly spend figures, and the performance ranges are operator claims that were not independently verified.

## 2. Alpha / Sauce

- **Native is a reading environment, so it sells through an advertorial.** Do not send native traffic directly to a VSL. Run it to a pre-landing page first, qualify the reader, then hand them to the video. He says sending direct to the VSL simply did not work in tests.
- **Keep 20% of traffic on a test, always.** The whole method is active traffic splitting against a control; you never stop testing angles and variations. The testing culture, he says, came from the copywriting side and was adapted to traffic.
- **Be careful what you admire.** In native, the first point of contact tires fastest, so optimization should start at the top of the funnel and work down, not the other way around.
- **Ad success is roughly 40% image, 40% headline, 20% description.** The description should be a three-line summary and preview that qualifies the click and matches the VSL, not a throwaway.
- **Get out of the herd.** Everyone using spy tools copies the same image with one word changed; watch what the herd does and present the opposite. That inversion is what restarted his offers when everyone else saturated.
- **Native depends on the offer more than the button-pushing.** A good offer converts even with mediocre creative; a bad offer cannot be rescued by native optimization.
- **Separate mobile and desktop, never Android and desktop.** Desktop clicks are more expensive; mobile is stable. He runs a mobile campaign and a desktop campaign rather than segmenting by operating system.
- **Renew the funnel, not just the ads.** His campaign that "died overnight" did so because they replicated one angle and never refreshed the first customer touchpoint.
- **The pre-landing page is high leverage.** Swapping the pre-landing layout and copy moved passage from the pre-landing to the VSL from about 13-17% to 40-50%.

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

1. Learn the network before spending; read the platform documentation and the tracker documentation, or hire someone to set it up, because configuring postbacks and pixels takes about a week from scratch; setup: a tracker such as Voluum or HeadTrack connected to the network via postback.
2. Wire the affiliate network to the tracker with the click ID and conversion ID so you can see every conversion end to end; setup: pixel plus postback for the offer.
3. Start by sending traffic to a pre-landing page, not the VSL; setup: an advertorial or text pre-landing that qualifies the reader and passes them to the video.
4. Build the ad in three parts image, headline, and description and make all three congruent with the offer and mechanism; setup: an image tied to the mechanism (for example, the substance or mechanism shown in the funnel).
5. Use the description as a three-line preview that summarizes the VSL promise and pushes qualification; setup: a short "expert reveals discovery" style preview.
6. Set a strong control, then run the test split; setup: keep about 20% of traffic on a new variable while the control keeps the rest.
7. Test in priority order from the top of the funnel down: pre-landing layout, headline, VSL headline, price, then autoplay; setup: one variable per test, measured against the control.
8. Because the top of the funnel tires first, refresh the highest-exposure touchpoint before the deeper ones; setup: a rotating set of pre-landing and headline variants.
9. Watch rival angles with a spy tool, then deliberately do the opposite; setup: AdPlex, Anstrex, or similar to see what the herd is running.
10. Launch two campaigns per offer one mobile, one desktop each with about six creatives spanning different angles; setup: separate campaigns, R$300 per day each.
11. Treat day one as the learning phase and judge on day two; pause any creative whose cost exceeds your acceptable initiation-checkout or CPA threshold; setup: a per-creative cost ceiling.
12. Scale in controlled steps at night, not midday: start R$300, hold a few days, then move to R$600, then R$900, then R$1,200 and R$1,500; setup: a configured run window (for example 6 a.m. to 10 p.m.) and changes made after the day ends.
13. Accept a modest CPA increase while scaling, then decide the next day whether to hold or trim; setup: a target CPA band and a daily review.
14. Watch the viewable click-through rate, not the raw click-through rate, because the platform rewards ads that make people stop for about three seconds and click; setup: the network's viewability metric as the primary creative signal.
15. Professionalize with a documented framework, a weekly performance report shared with the copy team, and a feedback loop between traffic and copy; setup: a reporting cadence that turns campaign data into new copy variations.

## 4. Tools & Stack

| Tool | Role |
| --- | --- |
| Taboola | Primary native network; the bulk of his spend |
| Outbrain | Second native network; about R$200k/month at peak |
| Voluum / HeadTrack | Traffic tracking, conversion attribution, postbacks |
| AdPlex / Anstrex | Spy tools to see competing native ads and angles |
| Google Display Network | Where he and Mateus first ran cheap-CPA nutra display |
| Pre-landing page builder | Advertorial/pre-VSL page that qualifies native readers |
| VSL host (Vurb-style) | Video sales page with autoplay testing capability |

## 5. Actionable Checklist

- [ ] Send native traffic to a pre-landing page, never straight to the VSL.
- [ ] Configure your tracker and postbacks before scaling anything.
- [ ] Keep 20% of traffic on a test and 80% on the control at all times.
- [ ] Build every ad as image, headline, and description, all congruent with the mechanism.
- [ ] Write the description as a three-line qualifying preview of the VSL.
- [ ] Test in top-down order: pre-landing layout, headline, VSL headline, price, autoplay.
- [ ] Refresh the top-of-funnel touchpoint first, because it tires fastest.
- [ ] Run a spy tool, then deliberately invert what the herd is doing.
- [ ] Launch separate mobile and desktop campaigns with about six creatives each.
- [ ] Pause creatives that breach your CPA ceiling on day two, then double the winner's budget.
- [ ] Scale in one step per day at night, never midday, from R$300 up through R$600, R$900, R$1,200, R$1,500.
- [ ] Track viewable click-through, and feed campaign results back to the copy team weekly.