OnX Backcountry is a GPS navigation app for outdoor enthusiasts, offering a wide range of features to help users plan and navigate their human-powered adventures.

Apple Search Ads represents a large portion of onX’ marketing mix, and the onX team was looking to increase its efficiency and effectiveness to be able to scale up this channel even further.

Phiture’s ASAi+ models were employed to optimize keyword bids across several campaigns. The models, trained on historic performance and proprietary market data, quickly provided onX with the desired increase in installs, whilst reducing their Cost-per-Acquisition (CPA). ASAi+ effectively managed the campaigns without the need for manual intervention.

Multiple separate campaigns were tested, resulting in a total 20% increase in installs and a 2.2% reduction in CPA within the first month.

The main campaign generated a 29% uplift in installs whilst reducing the CPA by 6.6%, improving campaign efficiencies and effectiveness significantly.

We focused on the first 4 weeks after launching ASAi+ to evaluate performance, this period provides the most accurate reflection of the model’s immediate impact, independent of seasonal effects.

foodspring aimed to enhance customer engagement and long-term loyalty by reducing its reliance on promotional emails and implementing personalized user experiences. Recognizing the primary obstacle to sustainable growth was the lack of a sophisticated automation strategy, their mission was to create a dynamic and compelling approach that would drive key metrics such as increased user interaction, retention, and overall revenue.

Taxonomy Deep Dive and Implementation

The project began with an in-depth analysis to categorize customer interactions for Braze’s automation. This process enabled precise tracking of various events and actions, setting the stage for personalized automation. Web SDK events were meticulously implemented through Google Tag Manager to help support this categorization.

Multi-channel Strategy and Automation

Together, Phiture and foodspring were able to develop a comprehensive multi-channel strategy incorporating emails, web pop-ups, and seamlessly integrating promotions with automation flows.

Personalization using Braze

Personalization was a key focus, utilizing Braze’s advanced features to enhance relevancy and efficiency. Tools such as Liquid Logic, Catalogs, Content Blocks, Webhooks, Promocodes, APIs, and Audience Sync were leveraged to create a highly personalized customer experience.

Achievement of Revenue Targets

Through our engagement, foodspring successfully met revised revenue targets for eight consecutive months, demonstrating the effectiveness of our strategies and collaboration.

Expansion of Live Flows

We expanded from tracking just two events to monitoring a broad array of events and attributes, building 25 live flows. This ensured customers received personalized, relevant communications, significantly boosting engagement capabilities.

Significant Revenue from Automations

Automations began contributing to 40% of foodspring’s revenue, highlighting the success of our engagement strategies. Notably, there was substantial growth in categories like Onboarding, which saw a 7.05% month-over-month increase since the start of the year.

Impressive Revenue Growth

From January to May 2024, foodspring generated €2.13 million through CRM activities, accounting for 20.31% of the company’s total revenue. This solid contribution underscores the overall business performance improvements driven by our strategies.

The primary goal of the six-week pilot program was to drive new installs for Found It!—targeting key markets like the US, UK, Canada, India, Japan, and Canada. The core objectives were:

• Automate the creation and testing of short descriptions and icons.
• Identify the highest-converting assets and apply them globally.
• Achieve significant increases in install conversion rates (CVR).

Phiture leveraged PressPlay’s GenAI-powered A/B testing capabilities to automate and accelerate ASO experimentation. Key features of the solution included:

• GenAI Asset Creation: Automatically generating hundreds of store listing assets like short descriptions, icons, and feature graphics, tailored to each app’s needs.
• Experiment Assistant: A self-learning system that manages, monitors, and kills non-performing experiments, freeing up test slots for more promising assets.
• Scalable Testing: Over 100 experiments were launched in just five weeks, with AI identifying winning assets and replicating them globally across multiple markets.

The pilot demonstrated strong results, particularly for the Found It! app, which saw a 20% global lift in installs, validated at a global scale. This uplift was driven by highly effective icon tests, which provided a statistically significant increase in conversion rates. The first successful icon test resulted in a 12% increase in installs over a period of 20 days. Following this, another icon variation led to an additional 8% increase in installs after 8 days. This incremental testing approach compounded the overall growth, leading to a significant boost in conversion rates, as each winning experiment built upon the results of the previous one, maximizing Found It!’s install rate.

• 100 experiments run in six weeks
• 600k projected annual additional installs
• 20% lift in global installs for Found It!

Based on the success of the pilot, Phiture scaled the A/B testing framework to new titles and continued to iterate on high-performing assets. The next phase includes:

1. Global Expansion: Applying the winning icons for Found It! to all markets.
2. New Titles: Expanding to new apps to further enhance testing capabilities, like Hexa Sort
3. GenAI and Automation: Continuing to evolve PressPlay’s GenAI technology to handle more complex asset creation.

For Hexa Sort specifically, Phiture utilized PressPlay’s fine-tuning capabilities to produce assets that are aligned with the original art style.
Read more about how we fine-tune our models for art styles and specific characters here.

Phiture’s PressPlay platform demonstrated its ability to rapidly drive app growth by automating A/B testing and scaling high-performing assets. With 100 experiments deployed in five weeks and a significant uplift in installs, this pilot provides a solid foundation for future testing and growth across new app titles.

Avakin Life, a highly successful mobile game with over 100 million downloads, approached Phiture with the challenge of driving app downloads and improving its conversion rate on the app stores.

The primary goal was to increase the speed and scale of testing experiments to enhance performance. However, several bottlenecks hindered this progress:

• Unique design: A large number of visual assets was required to run A/B tests at scale, each asset had to be created by a 3D artist to closely match Avakin Life’s unique visual style
• A/B testing automation: The limited number of experiments that could be run manually restricted optimization pace

PressPlay addressed these challenges by automating the creation, deployment, and analysis of A/B tests on the Google Play Store.

Leveraging GenAI, PressPlay generated hundreds of visual assets for testing without additional localization and creative resources.

The tool integrated seamlessly with the Google Play Console, managing the experiment backlog and setting up rules to schedule, apply, or stop experiments automatically. This allows over 50 concurrent experiments per app to be run at the same time, accelerating time-to-win.

Additionally, the platform continuously collected data from all experiments across portfolio apps, allowing for comprehensive analysis and optimization.

• PressPlay ran 110 experiments on Google Play in a 2-month period
• 19% conversion rate increase in the first month
• Another 32% conversion rate increase, after fine-tuning the AI model to Avakin Life’s visual style in the second month
• Total conversion rate uplift: 57% in two months

The fine-tuned AI model’s ability to consistently mimic Avakin Life’s characters and art style has opened new avenues for leveraging such models in the game’s broader marketing strategy. PressPlay’s self-learning system continues to evolve, promising even more efficient and effective ASO A/B testing automation in the future.

This case study illustrates the impactful combination of automation, AI innovation, and strategic application in App Store Optimization, paving the way for future growth and success in the app and mobile gaming industry.

Wildlife Studios faced the challenge of optimizing their A/B testing processes to keep up with the rapidly evolving mobile gaming market. The goal was to not only increase the speed and breadth of their A/B testing but also to ensure that each change contributed positively to app performance and user acquisition.

PressPlay, Phiture’s solution, was introduced to revolutionize Wildlife Studios’ A/B testing approach. Integrating Generative AI, PressPlay automated the creation, deployment, and analysis of A/B tests, making the process more efficient and data-driven.

The implementation targeted major titles, beginning with a focus on Sky Warriors, and later expanding to all major titles. Centralizing experiment data in Google Sheets, PressPlay provided Wildlife Studios with real-time updates and the ability to deploy experiments effortlessly.

PressPlay’s effectiveness was quickly evident:

• A total of 168 experiments were deployed in under 2 months
• 322 different variants (icons, short descriptions, feature graphics, screenshots) were tested, showcasing the depth and variety of the experiments.
• Notably, there were 7 significant experiment wins with GenAI.
• These wins contributed to an impressive 12 million projected installs, thanks to conversion rate (CVR) increases of up to 24.1% across winning experiments.

Wildlife Studios’ reaction to PressPlay was overwhelmingly positive. Matheus Magalhaes (Wildlife Studios), remarked, “The automation and intelligence of the system exceeded our expectations, establishing a new benchmark for efficiency in app store optimization.”

Matheus also highlighted the operational benefits: “PressPlay has significantly simplified the monitoring and deployment processes for our extensive experiment backlog. Managing thousands of experiments across our titles has become more streamlined and efficient, allowing us to focus on strategic decision-making and creative development.”

This enthusiastic feedback underscores the value and impact of PressPlay in transforming Wildlife Studios’ approach to app store optimization. The future roadmap driven by the collaboration of Wildlife and Phiture, focuses on leveraging these positive outcomes to further refine testing strategies and extend the technology’s application.

Example use cases and experiments in PressPlay — click to zoom in

The learning system and predictive analytics of PressPlay stood out:

• An Automated Experiment Assistant was built for each title, and trained on the app’s information, the competitive landscape and customer reviews.
• The Assistant is trained on all historical wins and losses so it better understands what works and what doesn’t
• Experiments are launched quickly and halted as soon as early predictive signals come in that a variant is losing.
• Its predictive capabilities on what copy or creative would win, improved over time, increasing the accuracy and success rate of future experiments.

Phiture’s PressPlay technology – developed by Abdulmajeed Alkattan and Moritz Daan – has set a new standard in mobile game publishing for Wildlife Studios, achieving 12 million additional installs in just two months. This case study illustrates the impactful combination of automation, AI innovation, and strategic application in app store optimization, paving the way for future growth and success in the app and mobile gaming industry.

Wildlife Studios is one of the 10 largest mobile gaming companies in the world.

Client: Wildlife Studios
Website: https://wildlifestudios.com
Services: PressPlay — Automated ASO Testing with GenAI