Success Story

Smadex was able to achieve great ROAS within the 4-week test period, focusing the strategy on LAT traffic. The deployment of a custom algorithm built specifically for the campaign was crucial for the increase in IPM, being able to gain very high CPMs, maintain lower CPI, and deliver high volume and great ROAS within the limited time frame.

• -51% CPI
• 17.8% ROAS D1
• 21% ROAS D3

On the verge of a new scenario coming post-IDFA, Socialpoint partnered with Smadex to do things differently and explore new bidding options. The experiment was set for a 4-week period where Smadex limited all bidding to LAT traffic. The clear goal was to acquire high-value users for Dragon City, with aggressive KPIs in cohort ROAS Day1 and Day3.

Once the campaign was live, Smadex’s algorithms started learning how to best reach the set goals for user acquisition. Through our optimization model
(Seed, Feed, Grow) apps were tested through our performance algorithms before reaching the Grow stage. By combining our exploration and pacing algorithms, Smadex’s team was able pay the smallest CPM possible at the beginning (precisely to be efficient at the exploration stage), and relying on our machine learning to find users to whom we’re willing to pay premium CPMs.

A breakthrough moment came 11 days into the campaign when Smadex deployed custom algorithms specifically designed and trained for Dragon City to scale the campaign with aggressive settings that optimized towards the clients’ CPI and ROAS goals. (as seen in the image).

For us at Smadex, this experiment meant to put our data science and iOS 14 working framework to the test. And so, show how our capabilities for contextual targeting were up to the challenge while:

Putting our optimization model to the test and efficiently finding great value users
Measuring the effectiveness of Smadex contextual ML algorithms to deliver high ROAS