User acquisition can be a make-it-or-break-it scenario for game makers. Attracting not just players, but users willing to make in-app purchases is a struggle faced by many developers and publishers hoping to scale up their operation, however the recent imposition of data privacy restrictions means that game companies worldwide have had to reevaluate their approach to advertising.

So how can a business still thrive whilst respecting user privacy? Working with a specialised Demand-Side Platform (DSP) can help game makers bring their work to the attention of new audiences, whilst analysing what works and what doesn’t in a way that complies with regulation.

In this guest post, LifeStreet CEO Levi Matkins discusses how working with a specialised DSP can help companies scale.

With iOS 14.5+ and the advent of new data privacy restrictions, the process of app growth has gotten more complicated. But, while acquiring high-quality players might have gotten more expensive and complex, mobile game engagement and playing session times are increasing across game app categories. This means marketers have an opportunity to thrive in the post-IDFA landscape if they focus their resources on acquiring high-quality, paying players from the start. To do this, partnering with specialised DSPs that have deep programmatic experience, historical machine learning models, and custom model testing capabilities is critical.

Below, we use insights from our work with mobile game developer, Small Giant, to illustrate the value of working with a specialised programmatic DSP in the data privacy age.

Small Giant’s focus and goals

Small Giant Games is a mobile game developer based in Helsinki, Finland. Along with two casual game titles, Small Giant has found success with a match-three, role-playing game titled Empires and Puzzles.

As a DSP partner to Small Giant, our team at LifeStreet supports their overall growth needs, as well as some of their region-specific launches. Below, we describe our work to support the expansion and reach of their titles.

How LifeStreet helped Small Giant

Along with supporting Small Giant’s overall user acquisition strategy, LifeStreet has also helped Small Giant grow their titles in the APAC region. To achieve Small Giant’s goals, our team built three custom targeting models to identify new, high-quality players and bid on them efficiently.

Geo-specific model

To improve buying in the APAC region, our team tested a model with training data isolating for APAC users. While this model did not outperform the global model we built for Small Giant’s campaign, it did allow us to try a new modelling approach to support its continued scaling up.

Advanced payer model

After initially scaling Small Giant’s campaign with a standard payer-prediction model (which predicted the likelihood that a new player would become a payer), Small Giant started passing our team more high-level data to sharpen our payer models. This included data on users that reached “Province 8” — an early indicator of quality users who would likely become paying players. We ingested this level-reached data into our payer model to align our predictions more closely with Small Giant’s quality goals. The outcome: we saw a lift in performance after we added the “Province 8” event to our models. This allowed us to further scale the campaign, achieving greater precision and cost efficiency in the process.

Retention model

Along with return on ad spend (ROAS) goals, Small Giant also had retention targets. Once the campaign gained traction and started achieving ROAS targets, LifeStreet’s CSM team moved forward on also hitting Small Giant’s retention goals. To do this, they layered and tested an additional retention prediction model onto their standard payer-prediction model. This model testing allowed us to improve the long-term performance of Small Giant’s growth campaign and unlock further scale.

How custom model testing works

The process for creating the above models required historical data and advanced filtering — especially when targeting players in specific countries (for example, users in Japan, Taiwan and South Korea).

● Signal testing: Ingesting and testing advertiser-specific signals is critical to driving performance. Each new signal is added to an existing model and iteratively tested against hold-out data. If the signal doesn’t move the needle, then it’s not used. However, if the signal has a positive impact, it’s added to the existing model and tested on live traffic.
● Machine learning: LifeStreet’s prediction engine uses machine learning to study historical data patterns and predict the probability of future desired outcomes. To build a payer-prediction model, for example, behaviour trends of users who have made IAPs are identified. These trends are then used to bid dynamically for impressions on subsequent users. Furthermore, our team has recently made enhancements to LifeStreet’s prediction engine that have enabled even more efficient testing and buildout of these models.

The outcome

LifeStreet’s custom model testing capabilities were able to support Small Giant’s region-specific and overall growth goals. According to Antti Paikkala, Head of Marketing at Small Giant, “Our collaboration with LifeStreet has been extremely fruitful for several years already. Their industry knowledge and experience has allowed us to regularly test new optimization and modelling strategies, allowing us to grow and maintain the consistency, scale and performance of our programmatic marketing campaigns.”

The value of a specialised DSP

As an agile, specialised DSP, LifeStreet was able to provide Small Giant with custom growth solutions better fit to their game apps and region-specific needs.

● Custom Solutions: Because LifeStreet is a small and independent DSP, we have the flexibility to create custom models for our clients. These custom models give them a competitive edge through differentiation (since they’re not using the same models as other game publishers). They’re also tailored specifically to fit clients’ media buying processes, events, and metrics — to better anticipate and react to their campaigns’ needs.

● Mobile gaming specific: With over 100,000 new apps released across the app stores every month and 25% of all mobile apps falling in the gaming category, mobile game app developers face a unique set of challenges when it comes to getting their apps noticed. This has compounded with the recent changes to Apple’s IDFA, with mobile game developers now facing a new set of restrictions when collecting and leveraging data to serve more precise and relevant advertising. LifeStreet is better positioned to address these challenges based on its history and experience in the mobile gaming space compared to traditional programmatic platforms.

● Transparency & control: No longer is programmatic advertising a “black box” of performance. Today, advertisers want to know the inner workings of their campaigns — including bidding strategies, auction dynamics and tracking — so they can understand the value of every ad dollar they spend. As an agile team, LifeStreet can provide a greater level of transparency and data granularity in its campaign management and reporting to meet these demands.

Conclusion

Using custom model testing, our team at Lifestreet was able to identify new, high-quality players and bid on them efficiently to meet Small Giant’s ROAS goals and expansion in the APAC region. Furthermore, we were able to layer and test an additional retention prediction model to improve Small Giant’s long-term player retention.

Small Giant’s story serves as a relevant use case for the custom solutions required to meet the growth challenges faced by game app developers in today’s increasingly competitive and complex data privacy landscape. With that said, partnering with specialised DSPs that have programmatic experience, historical machine learning models and custom model testing capabilities is critical.

Edited by Lewis Rees