User Acquisition is becoming more complex
As the User Acquisition (UA) team, you saw that CPM (Cost per Mille) of your campaigns in your top ad network significantly increased. You started to create new campaigns in other ad networks. After some time, the number of installs started to fall. You are not sure whether you should increase the bids in your ad groups since the bids are already around your D7 ARPU (Average Revenue Per User) values.
Now, you are in multiple ad networks with multiple campaigns and ad groups, the amount of bid decisions you have to make has reached a point where you don’t feel you have control over them anymore.
Or, your game is showing impressive results and you achieved a desired ROAS (Return on Ad Spend). You do not want to change anything in the campaign because you fear harm the algorithm behind it and diminish the returns. But are you sure that you are not being too conservative? How do you know you’re getting the highest profit possible?
Competition is rising
You are not alone. The number of decisions that have to be made to optimize marketing campaigns has exploded. Meanwhile, according to Sensor Tower, the user acquisition space in mobile gaming has become more competitive with number of installs increasing significantly from 2019 to 2020 in all game genres, hypercasual being the leader¹.
App discovery is a massive challenge, and now with the thousands of gaming apps to compete with, leaves no other choice but to invest in UA to draw attention to the game.
UA professionals have to use multiple ad networks to scale their games while keeping their acquisition costs as low as possible.
Too much work
However, managing creatives, ad sets and campaigns across multiple platforms is a significant challenge when done by only manual work.
UA teams have to find the optimal bid and budget values for their campaigns in multiple ad networks while adapting to the market changes to get the most of their marketing spend. At the same time, they have to create new target audiences and test new creatives to beat their competitors in the battle of attracting new players.
A new paradigm : Full Automation in UA
There is only one option to process data from different sources and actively manage marketing campaigns to achieve optimal performance. Companies have to automate their user acquisition strategy to keep the complex marketing decisions under control. There are 2 automation strategies that can save your valuable time and boost the UA performance: Rule-Based and AI-Based Automation.
Rule-Based Automation systems require a domain expert to implement ‘IF X happens THEN do Y’ type logic to relevant business problems. The executive continuously builds rule-sets to capture dynamics of the changing environment and simplify the business process.
In the user acquisition setting, a rule-set can be defined as follows: increase country bids and budgets %20 when ROAS is higher than %80 in Facebook and Unity Ads.
Rule-based automation is simple to implement and it can significantly increase efficiency if there are similar actions to perform on a regular basis. However, rule-based automation comes with its drawbacks.
- It is not suitable to behavior of different cohorts
In each network and campaign, you acquire users with different retention and ARPDAU behavior. Using one-size-fit-all rules leads to sub-optimal performance and significantly diminishes your returns in some adsets.
For instance, increasing the bid %20 might be a good decision for one cohort but it can be too high for another, even within the same campaign. UA folks have to create a complex set of rules to handle different behaviors.
- It is not adaptive to dynamic environments
The significant changes that platforms or ad networks make severely impact the rule-set mechanisms. For example, Apple reshaped the ad measurement and analytics ecosystem with App Tracking Transparency (ATT) Framework after IOS 14.5+. When Limit Ad Tracking (LAT) ratios are up, acquisition channels lose their consistency and rule-sets cannot adapt to high variance in performance.
- It is not adaptive to recent changes
When market changes, general trends in critical variables (eg. CPM, Conversion Rate) can go up or down. In addition, revenue metrics of recent cohorts can deteriorate. Thus, achieving the same ROAS goal or Revenue goal becomes infeasible in the new world. In this situation, UA teams have to change their parameters in their complex rule-set mechanisms and it can be frustrating to repeat the same process frequently.
When the number of games, networks and campaigns increases, it is much more difficult to handle user acquisition with a complex set of rules.
A better approach: AI-Based Automation
The more advanced solution to the problem is bringing AI into the game (pun intended). AI-based systems learn from historical data and build complex relationships between variables. To name a few use cases of using AI in UA:
- Sophisticated algorithms can learn from past data and predict new cohorts LTV (Lifetime Value) for Ad, In-App Purchase or Subscription Revenue models
- An advanced searching algorithm can find the optimum bid by an explore-exploit mechanism to get the most installs while reaching ROAS goals
- An allocation algorithm can distribute budgets across ad networks continuously to find most promising distribution channel
- Machine Learning models can detect recent changes in new players and suggest new decisions before rules and humans can notice
Since there is a huge amount of data from platforms and MMP’s, letting machine learning “learn” the rules is much more effective and efficient.
Advantages of AI-Based Automation:
- AI-Based optimization can produce customized recommendations for different audiences in different granularity. It eliminates building new rule-sets for each new ad set that comes with a different behavior.
- Bayesian Machine Learning works with uncertainty and creates recommendations with confidence intervals. It is easy to adapt variance in the ecosystem into the LTV or Bidding Algorithms.
- It is also simple to adapt to recent changes with Machine Learning. By continuously retraining models, any data shift in the environment is detected and recommendations become responsive to even hourly changes. For example, if ARPDAU of your cohorts in one adset falls in the last two days, LTV predictions are updated and new bid recommendations are served to achieve your target ROAS, long before humans or rule-based automation systems even notice the need for a change.
Getting to the optimal
The most challenging part of AI-Based Automation is that it requires a deep knowledge on Artificial Intelligence and a huge effort on technical implementation which makes it very difficult to build in-house with insufficient resources.
How can you easily move to AI-Based Automation?
A deep data science knowledge and user acquisition expertise is essential to build more advanced acquisition solutions to win in the highly competitive gaming ecosystem.
Fortunately, Growads is here to bring the performance of AI algorithms with ease of use. Built by expert Data Scientists and Machine Learning Engineers, Growads optimizes campaigns with its proprietary algorithms in LTV Prediction, Bidding and Budget Allocation.
We want to empower all UA and Growth Teams to acquire more profitable users with less effort with the power of AI.
To sum up;
- The complexity of UA operations is increasing rapidly with change being the only constant
- Predefined rule-based automations are not able to capture the various changes that occur in the industry, in the specific ad network or in the behavior of the cohorts
- AI-based tools, built by experts save the UA folks time and produce more optimal results with higher granularity