Big Data, Millennials, India and Netflix

…Everybody Lies or Just Video Viewers in India and China?

In Everybody Lies, Seth Stephens-Davidowitz takes massive evidence from Internet searches and throws it in our faces.   Big Data proves, he claims, how much we lie, not admitting our real preferences and predilections—our tendencies to violence, porn and prejudice, to mention some of what we conceal from traditional surveys.   

If only it was that simple.  Big Data, unfortunately, no matter how Big, also hides things.  It doesn’t speak for everyone.  And not always equally for those it does speak for.  We are not all equal big data users.  Some of us are obsessive compulsive searchers, others broad multi-topic searchers or very narrow searchers, and some can be infrequent searchers, even almost-never searchers.  Not to mention the full never searchers, many of who live in rural areas—both in the U.S. and around the world, especially in developing regions. 

Missing rural preferences cost Hilary Clinton the electoral-college vote.  In Silicon Valley it has caused poor predictions of how fast everyone will have a smartphone, an expensive one.  But as long as we get the Millennials right, the assumption goes, we’ll be o.k.  They’re using search and smartphones as fast as Big Data can count.  The rest is history.

Still, are Millennials in Silicon Valley the same as Millennials elsewhere?   Take China.  With a huge number of connected Millennials—well over 300 million of them—they are out-clicking and out-searching their peers anywhere on the globe.  Yet are China’s Millennials the same as those in other countries?  Mostly only children they may be looking for peer relations while other Millenials—at least those from bigger families—may be trying to escape their peers.   Do Millenials in China really search and click like we do? 

India and South Africa have higher shares of Millennials than China does but far fewer of them are connected and far more are unemployed—over 50% in South Africa, for example.  Millennials in these countries have more time for online search but less disposable income to buy the tablets they need to search.  And they often share the same tablet or feature phone with several other members of their large households or even across several households.

This brings up Netflix, which Stephens-Davidowitz highlights as one of the first to follow what we watch—not what we say we like to watch—in deciding what new movies to bring to our attention.  Yet even U.S. Millennials may share a Netflix account across two or more persons.  Not surprisingly then, Netflix has been known to prompt a romantic comedy addict to watch a violent action movie and vice versa. 

As Netflix grows around the world, following it’s 100 country expansion plan, it will face this prompting challenge much more, as households of 5 to 10 persons become the norm in its Big Data customer base.  In fairness to Netflix, it is for now trying to reach the high-end households in India but even these may be larger than their peers in the Valley.  Sharing, not just selfies, is the rule in global cyberspace, even for high-income households in a broad array of cultures and countries. 

In short everybody may be lying, if not in the same way.  It’s time to start looking at the differences—to break Big Data down.   Correlation doesn’t work when you have a mix of users, offset by intensity of use, gender, income, adoption stage, time of day or year, or location within and among countries.  Intensive users’ clicks outnumber those of infrequent users in the search world, but appealing to them may not bring in new types of users, who, worldwide, outnumber intensive ones 100 to one.