There are a few generalisations we can make when it comes to generating appealing artwork. For example, according to research done by Netflix, images that have expressive facial emotion that conveys the tone of the title do particularly well in making people watch videos. Using visible, recognizable characters results in more engagement, especially if you include the villains. Also, images containing more than three characters are less engaging.
However, despite those generalisations, it is also true that tastes vary greatly and what might make one person watch something could just as easily make another person avoid it, even if the content itself may be appealing to both. Video services are gradually putting more and more effort into personalising services in numerous ways, with personalised ads and content suggestions. Personalising the User Experience itself will make for much more engaged consumers and increased video consumption.
We have defined an innovation program for 2018, where we are continuing to investigate, together with customers, specific technologies and use cases with the aim of defining future media solutions. One of the topics we are actively evaluating is how AI can help increase user engagement.
To that end, we have been collaborating with Amazon Web Services (AWS) to develop a prototype based on our respective cloud products. By combining the deep learning-based image and video capabilities of Amazon RekognitionTM with our video experience platform, Accedo One™, our Content Management Solution, Accedo PublishTM and our A/B testing platform, Accedo Optimize™, we can dynamically create, test, and adapt artwork, based on a number of different criteria.
Each piece of content will have pre-defined engagement tags, such as with emotional triggers or high-profile cast members. We can then automatically render multiple content thumbnails to match the pre-defined tags. Thanks to Accedo Optimize™, we can determine the level of engagement amongst consumer profiles, and how that changes depending on the different artwork used. This can then be adapted so that the best choice of artwork is used for each user group.
The combination of the Accedo One™ platform and AWS solutions demonstrates how granular personalisation within video can be done in a scalable way.