How Leveraging ML Techniques Can Exceed Your App Marketing Goals

Aarki Machine Learning

There is no “one size fits all” in marketing. Aarki builds custom retargeting models for each campaign, using in-app activity data gathered before a user lapses, which is used to forecast user conversion and post engagement activity.

Machine learning (ML) involves analyzing data to uncover patterns. With the use of historical data, ML models can be used to predict future behavior. ML can be used to predict a specific user’s next action, including if they are likely to complete a conversion action. It enables advertising platforms to leverage data effectively and target the right audience for a specific campaign. ML’s capabilities allow advertisers to automate optimization and target users at scale because their systems automatically learn and improve without the need to be explicitly programmed.

Learn more about machine learning in our White Paper here.

The Importance of Machine Learning in Mobile Advertising

ML models are developed around goals, such as return on investment (ROI), cost per install (CPI), or other specific actions like registrations, purchases, or game level reached. Aarki’s bid optimization is crucial in a real-time bidding setting, as we must predict the important metrics from a particular ad impression. We convert these KPI predictions into a cost per thousand impressions (CPM) bid, to ensure we acquire the user at the right price to meet overall campaign objectives.

Each mobile programmatic campaign requires a made-to-fit, custom machine learning model. This model is designed to be continuously optimized to support your campaign KPIs. With ML, you will be able to have automated optimization and targeting at scale and ramp up your programmatic media buying.


Case Study

Lynx Games, headquartered in Seoul, reached out to Aarki to run a user acquisition campaign for their HighRoller Vegas Casino Slots app. The campaign’s objective was to widen its user base while generating a strong return on investment (ROI). Additionally, to meet and exceed their key performance indicators (KPIs).

Through Aarki’s large repository of data, machine learning expertise, and profound knowledge of strategies for casino slots apps, we were able to attain our client’s goals.

How Aarki Leveraged ML

We utilized Lynx’s first-party data, combined with our insights from previous campaigns, to ensure that we delivered data-driven and hyper-personalized ad creatives to the right target audience. Aarki’s ML process follows three phases: Learning, Scaling, and Optimizing.

We started off with the Learning phase, where we collected installs and impressions, which we fed into our algorithms. In the Learning phase, we aim to get a deep understanding of a user’s journey, which we use to inform our production model. New apps that go through this phase take up to four weeks, but it only took 6 days for this app.

The Scaling (or the Install Optimization) phase happened next, where our ML algorithms forecasted the probability of users to install the app and optimized for a stable cost per install (CPI).

On the 6th day, we launched the Optimization (or the Event Optimization) phase, where we deployed custom models, developed from all the data we gathered. This phase is crucial to deliver against campaign’s KPIs like ROI and CPI. The models improve their learnings as they collect more data. The analysis showed that the install rate moved closer to install probability, and the event rate approached the event probability.

Lynx - Results_resived

As a result, the ROI was consistently above the goal in the Event Optimization phase.

There were a couple of other factors that contributed to this campaign’s success. Read the full success story here to learn how we were able to get the ROI consistently above the goal in the Event Optimization phase.

Topics: Machine Learning