AUSTIN, TX: Netflix, the online streaming service, has moved significantly beyond using demographics to understand its users, having learned this information was "almost useless" as an indicator of behaviour.
Todd Yellin, Netflix's vp/product innovation, discussed this subject at South by Southwest (SXSW) 2015 in Austin, Texas.
"Everyone's instinct was, 'Yeah, if you find out their age and gender data, that's fantastic'. But what we learned is: it's almost useless," he said. (For more, including details of its A/B testing strategy, read Warc's exclusive report: How Netflix enhances the customer experience.)
"Because, here's a shocker for you, there are actually 19-year-old guys who watch 'Dance Moms', and there are 73-year-old women who are watching 'Breaking Bad' and 'Avengers'."
The importance of looking beyond such basic characteristics has been reinforced as Netflix's operations have grown increasingly international in scope. "Taste is becoming more global," said Yellin.
An alternative on which to base its strategy simply involves getting consumers to rate the shows and movies they watch.
"You can just ask people and have them rate things on a one-to-five star scale," Yellin said. "At the peak, we had over half of our members rating over 50 titles."
The issue with this tactic, however, is that people often "pretend" they like certain things – replicating the problems frequently experienced where studies are premised on reported behaviour.
"We've spent so many resources over so many years getting those billions of ratings, making an amazing algorithm that could predict how many stars someone's going to give something," said Yellin. "It's helpful, but secondary."
Through monitoring and gathering statistics across several years, Netflix has thus effectively been able to establish a hierarchy of data points.
"What we've learned over time is: it's not who they are in a superficial sense – like gender, age, even geography. It's not even what they tell you. It's what they do," Yellin said.
"This is how it works in the hierarchy; if you give us age and gender and geography, hit play on one title – hit play on 'Dance Moms', hit play on 'The Avengers', hit play on 'Orange is the New Black' – and we'll know more about you than all the superficial stuff."
Todd Yellin, Netflix's vp/product innovation, discussed this subject at South by Southwest (SXSW) 2015 in Austin, Texas.
"Everyone's instinct was, 'Yeah, if you find out their age and gender data, that's fantastic'. But what we learned is: it's almost useless," he said. (For more, including details of its A/B testing strategy, read Warc's exclusive report: How Netflix enhances the customer experience.)
"Because, here's a shocker for you, there are actually 19-year-old guys who watch 'Dance Moms', and there are 73-year-old women who are watching 'Breaking Bad' and 'Avengers'."
The importance of looking beyond such basic characteristics has been reinforced as Netflix's operations have grown increasingly international in scope. "Taste is becoming more global," said Yellin.
An alternative on which to base its strategy simply involves getting consumers to rate the shows and movies they watch.
"You can just ask people and have them rate things on a one-to-five star scale," Yellin said. "At the peak, we had over half of our members rating over 50 titles."
The issue with this tactic, however, is that people often "pretend" they like certain things – replicating the problems frequently experienced where studies are premised on reported behaviour.
"We've spent so many resources over so many years getting those billions of ratings, making an amazing algorithm that could predict how many stars someone's going to give something," said Yellin. "It's helpful, but secondary."
Through monitoring and gathering statistics across several years, Netflix has thus effectively been able to establish a hierarchy of data points.
"What we've learned over time is: it's not who they are in a superficial sense – like gender, age, even geography. It's not even what they tell you. It's what they do," Yellin said.
"This is how it works in the hierarchy; if you give us age and gender and geography, hit play on one title – hit play on 'Dance Moms', hit play on 'The Avengers', hit play on 'Orange is the New Black' – and we'll know more about you than all the superficial stuff."
Data sourced from Warc
