With best recommendations: On searching and finding in streaming offerings
The catalogs of streaming providers are constantly growing. Nevertheless, users get the feeling that they have already seen everything. Is something wrong there?
The agony of choice: Many users despair at the recommendation algorithms of streaming platforms.
(Image: pixinoo / Shutterstock.com)
A "huge, confusing department store where everyone finds something – the junk first, you have to search for the rarities" – this is how Oliver Kaever in Der Spiegel describes the streaming provider Netflix at the start of the last season of "Stranger Things". All those who are involved in the development of the recommendation system at Netflix and thus work to ensure that viewers receive recommendations that keep them on the streamer for as long as possible, may feel addressed.
Of course, this is not successful if the platform mainly presents junk that does not meet the viewers' taste. Many streaming subscribers know the feeling of having somehow already seen everything. They sift through endless recommendations and simply find nothing suitable. While in the past one would quickly flick through TV programs, zapping 2.0 with streamers often leads to frustration: instead of streaming, the user, annoyed, chooses another entertainment offering.
One third gives up after 5 minutes
According to a Nielsen survey from June 2023, a streaming user needs an average of 10.5 minutes to find a movie or series they want to watch. In ten minutes, any TV viewer could have flicked through the channel list up to channel 50, if they can still find programs of interest to them that far down the list.
On the other hand, almost one in three streaming users loses patience after just a few minutes. A study by the Royal Institute of Technology (KTH) in Stockholm from 2022 states that 30 percent abandon the search after a maximum of five minutes if they cannot find anything suitable on a streaming provider.
The curious thing is: the search time has increased. According to Nielsen, it was only 7.5 minutes in 2019. And this despite the fact that recommendation systems have become better and better over time, meaning they react more and more individually to the user's viewing habits.
Cold Start and Filter Bubbles
The days of purely rule-based systems, which, for example, only display the top ten most-watched movies or newly added series, are long gone. However, such rankings still play a major role – especially for new users, who are a blank slate for the provider.
However, this so-called cold start no longer lasts long these days. "Within a few seconds, the system can recognize whether, for example, a woman or a man is sitting in front of the TV," says Marco Hellberg, Managing Director of Canal+ Germany. The system tracks, for example, the time a user spends on a title cover.
The cold start is not the biggest problem for streaming providers like Canal+, which offers a subscription service in Austria and Switzerland and the IPTV platform "The Entertainment Hub" for network operators in Germany. "Sooner or later, a recommendation system's algorithm will suggest more or less similar content to a user," explains Hellberg. Existing usage patterns are reinforced, leading to a filter bubble. This increases the risk that the user will cancel the subscription.
Videos by heise
One in ten cancels due to overwhelm
"In the current market situation, customer retention is paramount," says Lisa Jäger of the strategy consultancy Simon-Kucher. Jäger is the author of the "Global Streaming Study 2025". According to the study, price and content are still the decisive purchase criteria for a streaming subscription. Whether the streamer also provides suitable recommendations plays a subordinate role in signing up for a subscription.
"However, in the usage situation, recommendations take on a whole new dimension," says Jäger. The KTH study concludes that a total of 80 percent of streamed hours are influenced by recommendations. "Content is king, but content that is not found is not king," Jäger sums up the importance of effective recommendation systems.
Poor recommendations also contribute to cancellation reasons. In the Simon-Kucher study, a quarter of those who want to spend less time streaming explain that new, exciting content is missing. Of those who have canceled their subscription, 24 percent cite the lack of content as the reason. But one in ten who canceled their subscription feels overwhelmed by the content offering.
"If, as a user, I don't perceive the value of a product because the recommendations are bad, the price quickly seems too high," says Jäger, relying on the results of her own study. The main reasons for cancellations are saving money (49 percent) and too high a price for the service provided (28 percent).