5 Ways Machine Learning Accelerates Mobile Video

 In Mobile Video Machine Learning KRAKEN, the “Birdman” Case study demonstrates video lift engagement powered by machine learning. In “5 Ways Machine Learning Accelerates Mobile Video”, we dive into why brands are embracing video as a key marketing and storytelling tool and how machine learning can be used to drive higher engagement.

The hard reality is video is STILL LINEAR.  Even so, some are attempting to make them interactive like Jack White’s Interactive Video that allows viewers to choose their own adventure.

While the majority of brand videos are still stuck in a 15s / 30s pre-roll with a force fed content model, we’re starting to see a clear migration to long form and sponsored content that’s not just an interruption but instead it IS the story. Video machine learning is new and millions of videos can benefit from programmatic visual control. Why machine learning? Marketers don’t care what algorithms you’re using they just want to see:

Mobile Video Machine Learing Birdman post

Case study on the movie trailer “Birdman” Click to play lift achieved 3000% using machine learning technology.

  • Revenue
  • Efficiency
  • Effectiveness

Publishers are looking to achieve high KPI’s in order to increase overall spend while the media buyer is looking to lower CPA, without increasing costs. Publishers are trying to increase inventory and get the most out of their customer’s engagement. Machine learning enables both parties to achieve their goal by impacting revenue, efficiency and effectiveness simultaneously.  With this technology publishers are empowered to keep the user video engagement high over significantly longer periods of time which is proving to be an invaluable tool that will become imperative to all successful video marketing efforts.

What Marketers want to see?

  • Viewability
  • Video watch time
  • Audio on or off
  • When did consumer stop watching
  • Was the video paused
Video Viewablity Across the Web

Google research finds only 53 percent of PC video advertising is viewable.

Gone are the days of simply tracking web page hits. A more sophisticated marketer has emerged where data is king. However, video distribution and analytics are complicated. Machine learning facilitates the systems ability to learn behavior and automatically adjust marketing efforts based on active feedback loops. This virtual neural network driven by human interaction with video content creates a meaningful data set providing the foundation for mobile video intelligence.

Programmatic Explosion

Machine Learning Mobile Video Birdman Split Test KRAKEN

Graph shows real-time A/B testing of static image and KRAKEN image driven by machine learning. Machine learning makes it possible to stabilize and achieve lift.

Programmatic targeting reached an all time high of sophistication with it’s own machine learning and big data approach. Companies like RockFuel, Turn and eXelate have all perfected audience based targeting with advanced machine learning methods of aggregating massive sums of data to ensure that the right content is placed in front of the right people at the right time. The following are examples of machine learning techniques being used to enhance content engagement levels.

1. Algorithmic learning is used to determine what demographic segment responds well with specific content (e.g. videos).

2. Identification of habitual responses to visual objects by region allows for higher confidence of consumer engagement with content.

3. The type of content greatly affects the reaction of a targeted segment. Machine learning can track the visual preference of the video segments to give brands and content creators a new level of understanding as to what an audience will find most appealing.

4. Machine Learning can predict audience consumption. Plotting audience behavior across video types creates a consumption map, which can be used to predict things like video placement and cycle times.

5.  Reduce video fatigue and increase engagement by rotation of static video images (thumbnails). Static starting images face image fatigue due to a lack of visual changes, color and motion alterations. Continuous and dynamic changes in a static video image will increase audience interest and result in higher click to play rates as well as completion rates.

Visual Programmatic

Netflix has the capability to “predict” what you would watch next based on past viewing habits. Information like show/movie title and genre are compiled to help select Netflix’s recommendations. These algorithms are an example of something that pulls from the surface level information vs actual content within the video.

Netflix-Wants-Personalized-Recommendations-Instead-of-Current-Interface-443094-2 Visual content marketing is a very powerful method of attracting and retaining customers. Building a content story arch is key to perpetuating engagement and video is the most effective means to accomplish this. Publishers that leverage their audience to tune the video will achieve higher levels of revenue on their existing assets.

How do you see machine learning impacting video in the further and what video KPIs do you track that aren’t on the list? Let us know in the comments!

Leave a Reply

Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title="" rel=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>