How Alarmy Uses 200% of Its Ad Data

Subtitle: Collaborating with the Data Team

eCPM, impressions, ad requests... If your team is thinking about ad monetization, you've probably wrestled with these metrics to figure out how to improve revenue. Everyone knows data is important, but since you can already view data through the consoles of each mediation and network, some of you might not have even considered collaborating directly with a data team. However, no matter how much data you can see on your own, direct collaboration with a data team can create an entirely different kind of synergy.

So today, I'd like to share how Alarmy has advanced its ad monetization through collaboration with our data team.‍

Ad Monetization x Data Team = An Independent Data Environment

The starting point of collaborating with the data team lies in building an independent data environment. As mentioned earlier, mediations and networks already provide consoles that let you view data through their respective dashboards, but as the number of these platforms increases, the resources required for data tracking multiply.

Also, when you analyze data in these isolated environments, you inevitably risk making incorrect data analyses or decisions based on them. However, since we're dealing with monetization, the side effects of a bad decision will immediately impact revenue.

As a solution to this, DelightRoom has already built a data warehouse and is operating an independent data environment for the Alarmy product (Reference: ETL vs ELT, What's Your Choice?), and we manage scattered ad data by incorporating it into the overall data warehouse architecture through a separate pipeline.

Through this, we were able to gather scattered ad data into one place, improving accessibility and taking ownership of our ad data. With consistent data integrity management and flexible scalability, we can stably support changing ad operations strategies.

‍

Areas You Can't See with Just Ad Metrics

The biggest advantage of such a well-built data environment is efficient data monitoring. As we have previously covered at DelightRoom (References: How Often Does Alarmy Get a Health Check-up (feat. Ad Metrics), How to Analyze Why Revenue Dropped‍), the centralized data warehouse makes it easy to analyze ad metrics by any dimension you want, such as by client, country, or ad placement. We've also been able to build a more appropriate and accurate monitoring system through statistical analysis of historical data.

What's particularly encouraging is the change we've experienced as our data warehouse operation period has lengthened. As the time since building the data warehouse has accumulated, we can now grasp annual cycles beyond simple weekly or monthly cycle analyses, making our judgment of outliers in the monitoring process even more sophisticated. Based on this accumulated data, we can remove daily noise and determine outliers for granular Ad Unit and Mediation Group metrics as well. This allows us to monitor by clearly distinguishing between issues that actually require a response and those that are just for reference.‍

‍Beyond the efficient analysis of ad metrics, another crucial benefit is being able to organically connect ad data with service/product data. As these two areas of data are integrated into one place via the data warehouse, we can comprehensively analyze the impact of ads on the actual service beyond just a simple ARPDAU metric. This is not simply looking at two sets of data side-by-side; it has been decisive in finding the delicate balance between ad revenue and user experience.

For example, we are now able to answer in-depth questions based on data, like "Will a newly added ad placement negatively affect user retention?" or "Is a user who used to visit a specific ad placement 10 times dropping to 5 times?" Furthermore, by turning these questions into metrics, we have raised the data resolution of both the ads and the product together, allowing us to continuously monitor for fluctuations.

Taking it a step further, we started logging and loading user behavior logs internally, as well as data like latency that mediations and networks don't provide. This latency data has played a key role in maximizing revenue while minimizing the degradation of user experience caused by ad loading. The integrated data warehouse built this way provides a seamless data analysis environment and enables much more sophisticated decision-making.

‍

New Seeds Revealed by Data

This improved data resolution has opened up unprecedented new possibilities in our ad monetization strategy. The most important thing in ad monetization is accumulating know-how through continuous experiments, and now we can conduct analyses that consider various variables like the seasonality and DAU of each placement in a complex way, going far beyond simple revenue data. Through this, we can judge the effectiveness of each experiment more accurately and employ differentiated monetization strategies for different periods.

Also, this integrated data environment naturally became fertile ground for discovering new hypotheses. By looking at various phenomena occurring in ad placements from both ad data and service data perspectives, deeper interpretations have become possible. For example, we were able to segment and analyze the Floor Price of our Waterfall setup and envision variable pricing based on that. In addition, beyond the direct experiment results shown by the mediation console, we've become able to grasp the impact of a specific ad experiment on overall user experience and long-term revenue. The insights derived from these in-depth analyses naturally serve as the foundation for our next experiments.

As the process of starting with analysis, setting hypotheses, conducting experiments, and gaining new insights repeats, our know-how in ad monetization is systematically piling up. This data-driven virtuous cycle acts as the driving force not only for short-term revenue increases but also for sustainable and stable growth.

‍

Up Next…

Do you have a little better feel for the kind of synergy that can be created through collaboration with the data team in the domain of ad monetization? Everyone knows the importance of data, but I also wanted to tell you that truly utilizing data well requires an immense amount of resources and careful thought.

I think this is probably the time when you're getting curious about specific examples of Alarmy's ad monetization. I'll be introducing more detailed stories in upcoming posts.

‍Originally published at https://daro.so.