In comparison, Netflix only appears to provide joy: the almost-unlimited array of high-quality material available on the service. The platform hosts more than one person could ever hope to watch. Unlike Facebook or Google, Netflix directly charges its users a subscription fee for the service, a more straightforward transaction than the data bargain that other services represent.

To many, Netflix appears more like a traditional media company, one that has merely leveraged the convenience of the internet, as opposed to the rest of the Faangs.

Loading

However, that analysis downplays one of Netflix's greatest assets. The company holds a wealth of data on its users, and constantly prods it to improve the service. Its "recommendation engine" has become the stuff of legend for its ability to determine what people might want to watch next: some 80 per cent of what users view on Netflix is based on what is recommended.

Data about its more than 100 million users are combined with machine-learning algorithms and A/B testing - in which different models are tried on different groups to determine what is effective. The Netflix interface - the home screen that greets users when they open the app - is constantly evolving.

Without the company's recommendation engine, many users would have no idea where to start. But Netflix's use of data extends beyond what many might understand.

Last year, engineers detailed how the service is personalised down to the artwork that users see when browsing titles. While the company used to test for the single artwork that generated the most views, in recent years it has begun targeting specific versions at users.

The film Good Will Hunting may be illustrated with artwork featuring the film's protagonist couple Matt Damon and Minnie Driver to viewers with a penchant for romantic titles; while a user who has watched several comedies might be shown Robin Williams. The company is fond of saying that unlike a television channel, which typically presents the same thing to every viewer, there are 100 million different versions of Netflix. The goal is to keep users watching. Netflix has found that it typically has a minute-and-a-half between a user logging on and losing interest.

"If we present that perfect image on your home page (and as they say: an image is worth a thousand words), then maybe, just maybe, you will give it a try," the Netflix engineers said. Is this manipulation? In contrast to the days when parents would worry about their children spending hours in front of the television, binge watching is now a national pastime, and many may find it a better use of their time than constantly refreshing their social media feeds.

But how far should this go? Reed Hastings, Netflix's chief executive, has often joked that sleep is the company's biggest competitor. "Think about it: when you watch a show from Netflix and you get addicted to it, you stay up late at night. We're competing with sleep, on the margin," he told investors last year. If Netflix starts to win that battle, there may be more concerns about its addictive properties.

The company's granular level of data collection has not gone unnoticed. It was labelled "creepy" last year after the Netflix Twitter account revealed that 53 people had watched A Christmas Prince, a festive romcom, for 18 days in a row. It also paid $US9 million ($11.6 million) to settle a privacy lawsuit in 2012 after a supposedly anonymous data set exposed users' viewing habits.

Netflix does not share this data with anyone, or use it for anything other than to improve the service, which means it cannot be accused of misusing users' information.

But it has been suggested that the company could introduce an advertising-supported version, especially in emerging markets, if its growth slows down. If that happens, or its data prowess goes further than people are comfortable with, Netflix may not ride the tech backlash as ably as it has done to date.

Telegraph, London