Contents
Streaming services have completely changed the user experience: most moviegoers prefer to watch premieres online. How do the largest platforms attract viewers and what technologies do they use to do this?
About the expert: Mikhail Bondar, Product Director of START video service.
The global streaming platform market has received a major boost during the pandemic, projected to reach an incredible $2027 billion by 842, although some analysts believe the trillion barrier will be surpassed. At the same time, between platforms, the number of which is constantly growing, fierce competition has unfolded for the very limited free time of viewers who are physically unable to embrace all the variety of content.
An overabundance of content and ever-increasing demands from subscribers are forcing video services to look for new ways to attract and retain an audience: to independently create films and series, improve the technological basis of services, and so on. The fight is for those statistical two hours that the average subscriber is willing to spend watching videos every day.
“Smart” analytics
The growing demands of the audience make it necessary to look for an individual approach to various social, demographic and age groups, which differ greatly both in their tastes and in the patterns of content consumption. And here new analytical tools come to the rescue, aimed at studying the interests of the viewer.
For example, Netflix, relying on analytics, not only forms a hyper-personalized issuance of recommended titles, but also uses the collected information in a content strategy. It is believed that the statistics collected by the platform were taken into account when creating the script and choosing the cast of the series “House of Cards”, which has gained immense popularity and received three prestigious Emmy awards. A few years later, a similar approach was taken to the development and filming of The Crown.
START also takes the collected data into account when working on new titles. For example, we understand which actors or acting tandems are popular with our users, we use them in new projects and in the continuation of franchises. We also analyze the response to certain topics from different audiences.
By studying the preferences of different audiences, Netflix expects to use this knowledge even to personalize trailers in the future. People of all ages, nationalities and tastes will see movie ads tailored to their preferences.
Viewers can choose content until something “hooks” them. For this, not only trailers are used, which are advertisements for the film and include its best episodes, but also previews of 15–20 seconds. During this time, you can determine how the film was shot and what kind of storytelling it has. It’s like opening a book in the middle and seeing if you like it or not. Services are not just showing content, they are finding new ways to captivate viewers and immerse them in the world of the movie.
What services know about us
Viewers often wonder what platforms know about them and who they share this information with. One of the key tasks of any streaming platform is to know the preferences of its client. Without this knowledge, it will not be possible to offer exactly the film that he wants to see. Therefore, services collect and analyze some of the user’s actions on the site or in the application. For example, a recommender system works based on the request history. The more data, the more accurately the platform will be able to hit your preferences.
Each user has come across an agreement on the processing of personal data. Based on this agreement, the services may process the name, user ID on the site and in social networks, contact details, photo, information about the language, gender, date of birth, address, city and payment details. This issue is regulated by the laws of many countries. In particular, Europe has stringent General Data Protection Regulation (GDPR) requirements.
In the START service, user data is stored for a strictly defined time, which is specified in the Privacy Policy. It should be noted that the conditions for the terms of storage of personal data, methods of their use are established and regulated by the legislation of the Russian Federation.
Content adaptation
Viewers belonging to different segments of the audience differ from each other not only in genre and taste preferences, but also in the very model of viewing video content. These patterns include both thoughtful viewing of a two-hour movie, and “jumping” between short movie clips, similar to television viewing, often interrupted by commercials.
Streaming platforms are looking to address these differences too, offering cuts, for example, of 30-second promo clips inspired by TikTok content, or specially filmed ultra-short content that is especially sought after by younger audiences.
For example, Netflix recently launched the Fast Laughs product. With short but meaningful snippets, the platform offers a new type of video consumption and helps the viewer to understand how wide the content offer of this service is, where everyone can find a way to enjoy the evening.
Adaptation of content to different audiences is carried out in other areas. For example, startup Flawless has developed an impressive artificial intelligence technology that allows you to save the original facial expressions of an actor. In doing so, she redraws his lips so that they correctly articulate the film’s dubbed translation. This will help improve the promotion of films in foreign markets.
In fact, we are talking about a type of deepfake technology that is becoming more widespread in the content industry. Warner Bros. used it to advertise the new movie “Remembrance”. A promo site visitor could upload a photo and receive a personalized trailer with himself “starring”.
An interesting, albeit controversial, experiment in the search for new forms of content was the creation of Netflix’s Bandersnatch series for the Black Mirror series. The non-linear narrative in it is completely built on the choice that the user makes at key points in the scenario, thereby approaching one of the possible endings. The high potential for repeat viewings here, however, is offset by the high cost of producing such works. Time will tell if they resonate with a paying audience.
Services are developing that seek to win the attention of narrow audience segments. These include, for example, the anime-focused service Crunchyroll, the children’s platform Toon Gogfles, which specializes in horror films Shudder, and many others. All of them offer their subscriptions and compete with the “giants” for the same two hours of viewer attention per day.
Artificial Intelligence
The abundance of content, even within a single platform, makes the quality of the recommender system extremely important. All video services are constantly looking for ways to improve it, since it is it that primarily affects the viewer retention rate.
Now most video services use recommendations based on the analysis of the history of watched films, genres, tags, lists of actors and crew members. This method is far from perfect and leads to platform visitors endlessly scrolling through the recommended list, unable to stop at anything specific.
The Amazon Prime recommender system can be considered a successful case. Her film library, which includes more than 20 thousand titles, is analyzed by AI and recommended taking into account many factors that are not obvious even to film critics. It is interesting to note that initially this system did not show very good results. But video service engineers found out that the neural network was trained on too narrow a set of examples, which included “timeless classics”. After the AI was retrained on more films, viewer retention was twice as high as any other algorithm previously used.
The future of streaming services
In the coming years, the streaming services industry will be influenced by several important technological factors: