In the last blog about Adaptive Timeout™, we said: “Adaptation is the key to evolution.” In nature, evolution is often a product of environmental changes. For example, climates get colder, and species evolve to handle these new temperatures. In many ways, evolution is the reaction to these external circumstances.
But what if the evolution wasn’t just a reaction, but created a cause? Within technology, we’ve seen that this is completely possible. Take for example streaming platforms. When a user first starts to use one, the platform doesn’t know what movies or series to recommend to the user; it just suggests the most popular movies or shows on its platform. As the user watches more content, the platform grows and adapts. Does the user like Romantic Comedy, Sci-Fi, or Drama? It learns this and starts to suggest the specific content the user would want to watch. Soon through this knowledge, the streaming platform is influencing what the user is watching because it has learned about the user. This is the power of Machine Learning and is what Adaptive Timeout brings to the Ad Tech industry: Intelligent Adaptation for an Improved Evolution.
Why Machine Learning Matters
Traditionally, publishers trying to figure out the correct value for their wrapper timeout face a difficult challenge. They have to take into consideration a tremendous number of factors which affect how they should select a timeout value. What type of device is the user on at that moment? How strong is his WiFi connection? Is the user even on WiFi or on a mobile network? Some of these factors are simply not possible to understand through manual, aggregate data analysis, because they are constantly changing. Publishers end up spending time on selecting a timeout value that ends up ultimately to be a suboptimal timeout value.
Optimise Header Bidding Through Adaptive Machine Learning
Adaptive Timeout leverages an adaptive machine learning algorithm to quickly and easily set the optimal timeout value. This algorithm learns and adapts from an individual user’s network conditions and behaviour to calculate and constantly adjust the timeout value. This means that the publisher’s timeout value will take into consideration important factors such as the user’s device type and dynamic variables such as the network speed.
With Adaptive Timeout, a publisher’s header bidding solution can leverage Machine Learning to optimise their wrapper timeout to:
- Recognise the User Device Type: Adaptive Timeout recognises and accounts for a user’s device type. This allows the machine learning algorithm to take into consideration different device processing and networking capabilities.
- Recognise the Network Strength: Adaptive Timeout recognises and accounts for the user’s network speed, e.g. Wi-Fi versus 3G. It also factors in the user’s current bandwidth usage, allowing the algorithm to adjust the wrapper timeout to account for the actual network connectivity of the specific user in real-time.
As we continue our series on adaptive machine learning, we’ll discuss how we’re bringing this technology to other areas of Ad Tech. In the meantime, to learn more about Adaptive Timeout or other innovations, please visit our Knowledge Base.