FEATURE5 October 2012

Surveys can be big data too

Data analytics

EMI Music’s David Boyle tells Brian Tarran how crunching data from a million surveys will reveal new consumer insight.


“In the data industry, surveys get a bad reputation,” says David Boyle, head of insight for EMI Music. “Data people think surveys are a little imprecise, they worry about quality, they’re not sure where the samples are pulled from and the number of interviews isn’t that big compared to the number of records they are used to dealing with in their databases.”

But survey data can be “big, it can be high-quality and it can really make a difference”, Boyle says. He set out to prove it with an initiative launched this July in which EMI invited 150 data scientists to pore over the data gathered from more than a million surveys conducted on EMI’s behalf by Lightspeed Research.

The first of these was fielded three years ago. Initially, EMI was sceptical – and with good reason. “The research data they had used in the past wasn’t really what they needed so they had to go on a journey and fall in love with research data and learn how to use it to make decisions.”

Competitive advantage

Boyle’s task was to instil within EMI a belief that survey research can be helpful if you ask the right people the right questions and make the results available to people in a way that works for them.

It took time, he says, but EMI staff were eventually won round. Since then the record company has surveyed music buyers in more than 30 countries. “We’ve researched most pieces of music that we have released,” says Boyle. “Certainly we research every artist, sales channel, product, pricing strategy and marketing campaign.”

For at least a year Boyle and colleagues have been toying with the idea of making the survey dataset more widely available to people outside the business, but only now did they feel the time was right. “There was a phase,” he says, “where we were very excited about what we were doing, where we felt that there was a competitive advantage there, and so we didn’t want to tell anyone what we were doing.

“Now, though, we’re so comfortable with this – the culture of using data alongside skills and judgment to make decisions is so widespread that it actually doesn’t matter if people know what we’re doing anymore; the data itself isn’t the advantage, the culture is, so that allows us to use the data in more creative ways.”

Digging deeper

EMI is working with Data Science London to make the million-interview dataset – or at least a subset of it – available to data scientists. These scientists were set a task at a recent hackathon to use the data to create an algorithm that could predict a listener’s level of appreciation for songs and artists.

“By having other people look at the data, analyse it and draw conclusions from it, we’ll learn more about what is in that data and about consumers,” Boyle says.