Why neural networks are the future of mobile gaming growth

Mobile games have traditionally turned to programmatic advertising partners for their effectiveness in achieving scale. But as gaming markets become more competitive, an acquisition-at-all-costs approach simply doesn’t cut it anymore. Mobile games now live or die by aggressive ROI targets that require users to play consistently, engage with ads, and spend on in-app purchases. They need an ad partner who can leverage data in more innovative and impactful ways to make it happen.

This is where neural network solutions like Liftoff’s Cortex play a key role. Cortex is the next evolution of Liftoff’s machine-learning platform powered by neural networks. Moving models to neural nets represents a game-changing expansion in the capacity and types of data signals models can harness. The new ML models can identify more complex patterns from larger datasets to make accurate predictions for specific outcomes. For gaming advertisers, this means connecting your game with audiences more likely to engage consistently and spend more on in-game purchases over time.

First-party data is your advantage — do more with it

In the past five years, programmatic advertisers have seen the popularization of real-time bidding (RTB). RTB ads offer advertisers greater transparency into the ad buying process, enabling them to precisely target their ideal audience by delivering a personalized ad experience. This real-time approach has also given DSPs (demand-side platforms) that invest in robust machine learning a significant advantage.

Game developers have a wealth of first-party data that provides a holistic profile of their audience beyond their demographic, including how they play and what motivates gamers to keep coming back. Unlike traditional targeting, which relies on segmenting audiences based on characteristics like gender, region, or age, advanced machine learning solutions leverage historical data such as session time, ads watched, previous purchases, and others to model current and future user behavior.

Cortex, for example, combines first-party attribution data with exclusive data from the Vungle SDK and GameRefinery to predict how likely users are to install and continue to engage with a game or make purchases post-install. In real-time, it selects higher-value users to target.

Neural network solutions help games grow sustainably

For many mobile games, return on ad spend (ROAS) is still the standard KPI for a successful campaign. To maximize value from ad spend, mobile games need to find and nurture whales and high spenders. Deep learning, powered by neural networks, is better at spotting these quality users than traditional machine learning running on linear and logistic regression models.

For example, imagine a user who spends a lot on a live event but does not spend again over the next 60 days. Logistic regression models with limited memory and operating on binary divisions might misidentify the user as a low match if they can only separate users who spend from users who don’t spend. With better memory and the capacity to uncover more complex patterns, neural network models like Cortex can recognize a high-value user who spends periodically but spends big.

"The goal is a more personalized ad experience that delivers both performance for advertisers and a more relevant, non-disruptive experience for users"

Faster innovation for better ad creatives that build a connection

Of course, an essential part of growing a game is engaging and nurturing quality users once you have their attention. Superior, user-centric ad creative shows off what your game can do and sparks a connection with your user that can be the start, or rekindling, of a lifelong relationship.

Generally, the key to creative innovation is iterating on existing formats for more personalized forms of engagement based on user behavior. Neural net models accelerate creative testing by running more experiments faster, making it easier to identify winning ad formats. Liftoff now offers exclusive dynamic ad formats tested on Cortex that change depending on user behavior. Our new page-turner format introduces a different playable if the user does not engage with the playable offered. We were able to develop the ad format quickly thanks to accelerated testing on Cortex. In our testing, page-turners increase impression-to-install (ITI) by up to 9.5% and lower cost-per-install (CPI) by 7.5%.

Combined with behavior modeling, advanced ML solutions like Cortex can also leverage motivational frameworks to map what drives audiences to convert. Integrated with data from GameRefinery, Cortex can improve gaming ad creative performance by personalizing delivery, helping advertisers to match users with an ad that taps into what specifically motivates them.

Ultimately, the goal is a more personalized ad experience that delivers both performance for advertisers and a more relevant, non-disruptive experience for users.

Navigating transparency with neural networks

Another challenge gaming marketers face is privacy. Years after Apple rolled out its App Tracking Transparency (ATT) framework, advertisers on iOS are still struggling to work without device-level identifiers. Ironically, less access to device-level data actually increases the impact of neural network models. By uncovering complex patterns from historical data, neural networks can optimize for outcomes based on user behavior, moving models away from heavy reliance on single identifiers.

Cortex is already helping advertisers hit their targets

At Liftoff, we’re excited to see what Cortex can help our gaming partners achieve. On average, across our platform, we’ve seen 23% lower CPIs, 21% lower CPAs (cost-per-acquisitions), and 16% higher ROAS for our partners, and we expect performance to continue to improve.

Our partners are also sharing Cortex's positive impact on their business. Seokyung Lee, marketing manager at Playlinks said: “We are satisfied by working with Liftoff as a trusted partner. With their strong ML logic, Cortex, they have shown strong ROAS results compared to other networks and the proportion of new users is almost 80% on iOS.”

Performance marketing managers at Krafton, Sangbum Kim & Kevin Youn, shared: “By launching user acquisition campaigns using Cortex that aligned with BGMI's [Battlegrounds Mobile India] marketing goals, we successfully expanded our focus to purchase and retention optimization, leading to increased overall campaign ROAS and retention.”

Curious to learn more? You can read more about what Cortex can do here or contact Liftoff directly.