What does housewives’ porn consumption have to do with your startup? It’s in the data

Melvin Sumapung works with corporate strategy and business development at big data and analytics firm Dattabot (previously Mediatrac). He’s standing on the product stage at Tech in Asia Jakarta 2016, and he’s talking to us about porn.

He’s talking to us about porn.

“Who knows that there are basically a lot of desperate housewives in Indonesia?” he asks. A slide in his presentation shows all four iconic Desperate Housewives actresses, along with the result of a survey of 50 to 60 million people: “We find that, counterintuitively, housewives watch mobile porn a lot more.” Welcome to the wizardry of data analysis.

With magician-like swiftness, Melvin pulls out other quirky facts during his presentation. For example, a great indicator of people who are most trustworthy with credit are people who fill out the “optional” column on their debt forms. The Chinese horoscope tends to be a good predictor of farmers’ productivity.

No one expects the porn slide.

Given enough data points, one can pull information on a number of trends and relationships. But how do you pull useful data when there’s so much out there?

See: GE partners with Indonesian big data startup to make unforeseen blackouts history

Hurdles

The product stage is so packed that I can only manage to cram myself into the back of the room, where I and other standing individuals stand nearly shoulder-to-shoulder. Melvin says that good big data insights require connectivity, intelligence, and integration. Indonesia faces challenges there because internet penetration in the country is only 34 percent and the big data industry in the country is quite young.

Indonesia kicked off this year with an engineering shortage.

“In order for you to get a lot of data to be turned into insights, you need to overcome these [technological challenges]. There are a lot of data formats, structures, sources,” Melvin tells the crowd. “Preparing them is very time-consuming and complex.” Even if one executes that process without a hitch, there’s still the problem of communicating the data in a way that people will understand.

Solutions