Why TV needs data artists, not data scientists

Taken on its own, data isn’t a silver bullet, writes Genius Digital’s Giles Cottle – it takes a special type of artist to truly make it work

I loved Dominic Mills’ article about artists and data, and the London exhibition that inspired it. I particularly loved the decidedly analogue take on big data that Giorgia Lupi and Stefanie Posavec’s hand-drawn postcard data visualisations provided. And it’s hard to disagree with Jonathan Harris’ “Data Will Help Us” – a manifesto of, as he puts it, the “promises and perils of big data”, that closes the show.

In fact, I found myself agreeing with almost all of Mills’ points which, given the article’s title – Artists vs data…and art wins – and my line of work, may seem counter-intuitive. But we think there is a much closer relationship between the worlds of art and science, at least from a data perspective, than meets the eye.

This isn’t always obvious in TV-land. Take the the actual creation of the TV shows that we all know, watch and love. This has always been a polarising topic in the TV art vs. science debate, if you want to call it that. There’s something intangible and magical about what makes some programmes good, or some great – or why some people love shows, and some don’t.

I’ve never been able to get into The Leftovers on Sky Atlantic – yet I love most HBO output, Tom Perrotta – who wrote the books on which the show is based – and everything that Damon Lindeloff has put his name to (Prometheus, Lost, Star Trek).

Netflix has been long been flying the flag for the use of data in its commissioning process. Its use of its famous (or infamous) algorithms are well known, and its use of data is, it says, a big factor in its success to date.

But even this most scientific of attitudes still has art at its heart. Broadly speaking, Netflix’s approach is not to look at what I’m watching, and then try and commission or create programmes that it thinks I’ll like.
[advert position=”left”]
It takes pitches from commissioners, and then uses data to decide whether enough of its subscribers are likely to be interested in it for it to be worth it investing in the programme.

There’s a subtle distinction here. It isn’t using data to replace the creative process – it’s using data to better position things that come about as a result of the creative process. The great idea still comes first.

Art plays a part in some of the more mechanical aspects of the use of data in TV, too. Take the humble art of segmenting your audience, which has been “big data-ified” in recent years. Once upon a time, a customer segmentation in TV-land was largely based on an operator’s CRM, or any primary research on a sample of its customer base that it could get its hands on.

The advent of return-path data (or “big data”, if you must) has turned this on its head. An operator can now know exactly what all of its customers have watched, and when, and how. It’s an amazing wealth of knowledge to draw upon when it comes to segmenting your audience. (I’ve even heard people question the need for an exercise like a customer segmentation – but that’s for another day).

But there’s still an awful lot of “art” that goes with this science. Very simply, data on its own isn’t a silver bullet to answer all of your problems. And ultimately, the value of data does not lie in how many ones and zeros you have sitting in a data centre somewhere in the middle of the desert. It’s how you use data to improve your business – be that to gain more customers, stop your existing customers leaving, sell more, and so on.

And here’s where the art part comes in. Presented with the same item, Van Gogh, Dali and Picasso would all paint it completely differently. Likewise, presented with the exact same set of data, three data science teams would likely come up with different views on what the data said, and what to do about it (although one hopes they’d come to a closer consensus than Vincent, Salvador and Pablo would).

Yet there are so many reasons – people’s viewpoints, bias, experience and personalities – that make this the case. And that’s before we even get into instinct – that wonderfully intangible element that make us say that even if the data says X, we know that the right is answer is Y.

For this reason, we like to think of our data science team as data artists. We haven’t quite got as far as getting them to put that on their business cards yet, but we do think that “data science” doesn’t quite capture the essence of what they do.

If anyone has a better idea, then answers on a (hand-drawn) postcard, please.

Giles Cottle is director of consumer insight products at Genius Digital