Brazil remains one of the most active markets for crash games. But this market is very overheated because of rockets, jets, and space. That’s why we were looking for offers that don’t sell old visuals in a new wrapper, but give something simple, lively, and close to the Brazilian audience.
Hello! We are the NNN team, and this is our case study on driving traffic to the crash game Piggy Tap by Onlyplay. Let us tell you how we used AI-creo and FB Ads to fill the first test with only $1000 and have already made a profit.
We won’t lie – we learned about the offer from someone else’s case. Then we went to check how it could catch our CEO’s attention. What we liked:
As a result, everything came together, and we took the offer to work. The first tests were launched as a simple hypothesis testing, but we immediately came out in the black. But first things first.
We decided to build the test around AI creativity to save time and run more variations. We used a classic set of neurons: GPT, MidJourney, and Runway. The voice acting was run through ElevenLabs, which has a good, lively Brazilian Portuguese. But still, we always give the final file to a native speaker for verification.
We relied on three things in terms of approaches:
During the evening, we collected a bunch of creatives, plus a few more options that the guys from Onlyplay finalized for us – all of this was put into the first split.
We didn’t reinvent the wheel here – after clicking, the user is first taken to a short pre-land with Piggy Tap, and from there, they go to the casino. We worked only with mobile traffic – the desktop was cut off immediately: The PC audience in Brazil is worse at instant-win and simply blurs the result.
We threw $1000 at the test – three $330 campaigns, without any “overclocking” or complicated settings. We just wanted to test the hypotheses on creo. We poured purely through FB Ads, targeting Tier-2 cities in Brazil – Fortaleza, Recife, Belo Horizonte. The auction is cheaper there and the audience is easier to catch. And here’s what we ended up with:
The best performing dynamic videos were those with a girl tapping Piggy Tap followed by a win and emotions.
As for static videos, the best ones are the game demo and the initial bonus for new players. Creatives with national elements are very cool depending on the country of the gulf. So don’t be afraid to experiment. Good luck and high profits to everyone.
As for the conclusions on tests and hypotheses. We expected more from memes and classic gameplay. However, it turned out that the stories where the player loses something, takes risks, makes the last tap are more engaging, especially in the POV format.
The second conclusion is that you can safely start with neural networks and $300 if you understand what you want to test. We ran 15 hypotheses in one evening, and three of them pushed the envelope better than studio creos.
By the end of the tests, our $1000 turned into $2100 rhubarb, plus it became clear what to build the scale on. And how we rolled out this link further from $1000 to $5k, we will tell you in the next case study.