Success Story

The client’s main cost per subscription goals were achieved by every optimization phase performed by Smadex, showing a gradual decline in CPA since implementing Flexible Mapping.

• -56% Cost per subscription
• +320% Growth in new subscriptions
• -48% Cost per install
• +222% New installs

By meticulously applying Flexible Mapping phases, Smadex was able to reach and surpass campaign goals thanks to all the data gathered from SKAd campaigns.

An American subscription video on-demand service operating worldwide and focused on live streaming and original content was looking for a partner able to deliver results with SKAd.

The introduction of SKAd by Apple put in motion a new mobile programmatic framework marked by data receipt delays, loss of user-level data, and reduced visibility of post-install events. Even though SKAd does provide some information, it is not useful enough for campaign optimization.

This new scenario presented a challenge for the client. They needed to find a trustworthy partner that could deliver their marketing budgets while being able to optimize towards SKAdNetwork events, keeping costs per subscription on goal, even with the lack of information SKAd entails.

Smadex’s Flexible Mapping is a tool that allows advertisers to map a wide range of variables, such as creatives, ad sizes, inventory bundles, and exchanges (SSPs) by utilizing the 100 available campaign_IDs SKAd provides, without uniquely identifying the user and complying with Apple’s privacy policies.

With Flexible Mapping, advertisers can surpass the limitations brought by SKAdNetwork, identify the combination of variables that get the best performance and use this information to optimize their campaigns. The best and only way to make sure the client’s budgets were delivered in full, while maintaining performance, was to use Smadex’s Flexible Mapping solution.

The first step was to use Flexible Mapping at the exchange level to identify the best-performing exchanges. Taking into account the 24-hour to 5-day delay SKAdNetwork events come with, Smadex tracked the data during one month to make sure that there were enough learnings for automatic optimization.

There was no time to lose, and Smadex’s team saw an opportunity in using this time to lay the groundwork for campaign optimization:

1. First, by analyzing data gathered from Flexible Mapping applied to similar verticals in previous campaigns. Information about exchanges, inventories, and ad sizes would provide insights on which ones should be scaled and which should be removed from the client’s campaign targeting.
2. Applying contextual targeting by setting up a specific targeting strategy, a custom list of inventory, and historically successful inventory from similar verticals.

By doing this preliminary work, the campaign started with precise targeting, and performance was on track from the get-go. Most of the inventories quickly started receiving several installs per day, which allowed them to pass SKAd privacy thresholds and send Smadex information about subscriptions, making post-install performance trackable and optimizable.

This information was fed to Smadex’s machine learning algorithm to learn so it would be capable of performing automatic optimizations.

Once the first mapping period was completed and the data was in, the team performed optimizations, scaling specific exchanges that showed the best performance. This information was transferred to the algorithm, and the second stage began: removing Flexible Mapping in order to cross privacy thresholds faster and receive all the post-install events from SKAd.

Apple’s privacy thresholds present one of the biggest challenges for marketers to identify and learn what’s working and what’s not from their campaigns. Privacy thresholds mean that if the number of app installs per campaign ID falls below a number set by Apple, there aren’t enough installs to disguise the user’s identity, causing Apple to censor the data received. Since Flexible Mapping uses more Campaign_id values to map different variables, until the optimization phase is completed, it’s harder to reach the thresholds.

As the campaign matured, one exchange developed poor performance over time as no inventory coming from it was passing the privacy thresholds. Once Smadex’s algorithm learned this, it started spending less in this exchange until it was eventually removed from the campaign. Allocating that budget elsewhere allowed the campaign to gain improvements in performance.