By CTO & Co-Founder, Amit Attias
As we move forward, real-time data analysis will become more and more integral to creative personalization, fraud prevention, A/B testing, gaming, and much more. Companies that have adopted real-time ML technology have seen major lifts in their profit margins, and profits drive technological integration.
In programmatic advertising, the DSPs (Demand Side Platforms) and SSPs (Supply Side Platforms) that have more instantaneous access to data in order to make decisions during RTB (Real-Time Bidding) will compete better. This is forcing real-time data to the forefront of mobile advertisers needs.
Bigabid doesn’t like to “wait and see”. In our commitment to staying on the leading-edge of programmatic ML technology, we have actively invested in real-time data analysis to optimize our performance for both user acquisition and retargeting platforms. In this brief article, I will touch on how we apply real-time data analysis to calculating sessions, understanding interruptions, and how, when combined with Bigabid’s Deep Categories, we optimize performance in ways other DSPs can’t.
How Sessions and Interruptions are Defined in Mobile Advertising
Sessions refer to the activity and set of pages viewed by a given user on a specific app, group of apps, and category of apps, on a certain device without interruption. Bigabid uniquely calculates these session features in real-time and adjusts our pricing accordingly for optimal performance.
In order to calculate a session, we first need to better understand interruptions, as a session should be without interruptions. Exchanges use the same global interruption times for all categories. Other DSPs adopt these global session length signals from the exchanges. At Bigabid, we discovered that interruption times were very different between app categories, and moreover, they were very different between Bigabid’s proprietary Deep Categories.
To take a deep dive into Bigabid’s Deep Categories.
It’s important to note that just like Deep Categories, Bigabid’s real-time data analysis of session features is contextual, with the only exception of cross-app session behavior we use on Android. Being contextual, our real-time session feature data is primed for Apple iOS 14 changes and post-IDFA effectiveness.
For example, if we were to reward users on a puzzle game session to watch a video ad, some users go for the reward and others don’t. We can see if a given user is tolerant of ads, ignores ads, or even stops using the app. We can then relate this user behavior to their session data, find correlations, and enhance. All of this is done with zero deterministic data and is therefore perched for Apple iOS 14.
A Closer Look at Session Interruptions and Depth
Depending on the app, as determined by Bigabid’s Deep Categories, an interruption in a trivia game with set time limits could be as low as 30 seconds, on a quest game, where one can pause, it could be two hours, on a multiplayer game, where players are waiting, it could be less than 30 seconds. It’s very important to be in real-time so we can understand the session as it’s going. We need to know when they came back in real-time, and we need to know the depth the user is inside the session in real-time.
We define session depth in time measurements (e.g. 3 minutes in), impressions (e.g. 20 impressions in), apps (e.g. 3rd app in this session)– if we are looking at cross-app behavior. Understanding these features in real-time dramatically improves our pricing, and therefore, performance. We know when to present an ad and how to price it nearly instantaneously to reach high LTV users at the right time.
Discover more about Targeting High LTV Users.
It’s also important to note that session features aren’t limited to gaming apps alone. If we were using session feature data for a dating app, we could identify the “patient” ones who have time to waste versus users who are less “patient” and will pay for results. In this way, dating app behavior is very similar to gaming app behavior. Real-time session feature data would optimize performance in both spaces equally as well.
Power of Real-Time Data Analysis + Deep Categories
By implementing Deep Categories– Bigabid uniquely delineates 1000+ app categories instead of the paltry few dozen other platforms use– we learn that a match-3 player behaves differently than a tap-3 player. We even understand that a home-design match-3 player behaves differently than a classic match-3 player. They won’t be interested in the same ad.
Using sessions combined with Deep Categories, we also understand that a home-design match-3 player with a 20-minute average session length is different from a home-design match-3 player with a 40-minute average session length. They react differently to ads, designs, apps, etc. Both app Deep categories and sessions help us understand users and audiences in real-time to determine how to best engage them.
This ties into how session features can affect ad formats. Because some session data points are split per ad format, some of the algorithms we developed can calculate different weights for them according to their Deep Category (e.g. value of a user seeing 10 banner ads vs 1 rewarded). Session and Deep Category data enhances our creative personalization process in real-time to not only show the right ad at the right time to the right user in the right mindset, but it can also help determine exactly how that ad should be displayed given its creative variations.
Discover more about our Creative Personalization process.
Positive Feedback Loop Between Your ROAS and Our Technologies
As you can see, real-time data analysis adds an exciting performance-enhancing element to user acquisition. Targeting in real-time based on session features and Deep Categories greatly improves our pricing, and therefore, your ROAS.
Bigabid’s devotion to the constant evolution and refinement of our technologies pushes us to create faster and smarter ML. Fortunately, that devotion returns good business, which in turn, drives our technology to push the envelope even further. It’s a positive feedback loop that we enjoy being part of. If you’re interested in seeing how real-time data analysis can go to work for your app or simply discuss its unique, cutting-edge inclusion into our technologies, please click the “contact us” button below.