Dynamic Decisioning Using Real-Time Machine Learning – The 3 Most Important Types of Features

Dynamic decisioning is a powerful technique for optimizing user journeys. Choosing what to show an individual user in order to incentivize the right action, including actions like paywall conversions and registration conversions, can be one of the most important drivers of revenues for B2C companies.

Machine learning enables businesses to adapt each individual user’s journeys to both changing user mindsets and changing market conditions. Layering in real-time machine learning techniques enables machine learning algorithms to leverage the latest user information which often reflects their current mindset. At Vidora, we combine both real-time in-session features with batch features computed over a longer historical time period in order to drive the optimal user journey.

Real-Time Machine Learning Features

There are generally three types of features which are leveraged for Vidora’s real-time machine learning:

1. Contextual – These features reflect data on the current user session including features like Time of Day and Device. For instance, a user may be more predisposed to sign-up for a subscription on their desktop at night than on their morning commute on their mobile device.
2. Behavioral – These features reflect information on the current user session including features like The Session Length or the Number of Pageviews. Details on the current session behaviors are often important signals for determining the state of mind of a user as they think about buying or converting.
3. Custom – Vidora also provides the ability to leverage custom features designed by the business and typically unique to that business. Examples here include the Number of Views of Nike Promotion Content or the Total Value of the Checkout Cart. Custom features provide a flexible mechanism for businesses to add in their own insights into what is most important for user journey optimization.

Dynamic decisioning which leverages an adaptive machine learning system can drive an immediate meaningful impact to your business. Our goals with real-time machine learning are to provide a flexible framework which enables your business to leverage both real-time and batch signals in order to ensure that you are maximizing conversions across your user base.

If you are interested in learning more, please let us know at info@vidora.com and we can schedule time with our integrations or data teams!