What does ‘big data’ mean for journalists?
edits this blog and produces BBC Business films @chblm
There’s a lot of talk about ‘big data’ – the vast quantities of information that can be captured and processed by companies from their billions of online interactions with customers.
It's mind-boggling. But what can you really do with big data?
Last night’s First Tuesday meeting offered a new thought, bringing the idea more into the realm of the practical. All the speakers agreed that, instead of being wowed by the huge numbers and fancy graphics - to date the hallmark of big data discussions -, if we want to harness all that information, we need to mine it for useful details, rather than sitting back and admiring the (very) big picture.
In other words, it isn’t the billions or trillions that count but little slithers of big data. That’s where businesses can exploit the new science and technology.
Of course that didn’t stop Richard Hammell of Deloitte Analytics from dropping in the obligatory drop-dead stat: that the amount of information produced annually is now the equivalent of 5,000 trillion books, which would stretch from here to Pluto (or was it Mickey Mouse?).
Not to be outdone, Libby Robinson, from the mobile division of advertising agency M&C Saatchi, said that last year, 597 petabytes of information were produced online per month, which is eight times the size of the whole internet in the year 2000.
I think that’s what she said.
Anyway, it’s a lot.
The third speaker, Neil Morgan, digital marketing director at Adobe, explained why his company has decided to shift most of its marketing budget online, to use the power of big numbers to raise the efficiency of its sales operation.
He showed two pieces of artwork promoting an Adobe product on Facebook and asked his audience to guess which one produced more click-throughs. It was the more visually exciting one. But it turned out that the one with more information actually led to more sales.
By experimenting with different designs, the company can finesse its message, tracking precisely what works and what doesn’t.
It gives the lie to the old joke about advertising: ‘I know half the money I spend on advertising is wasted: the problem is, I don’t know which half.’ Now you know which half. Except it won’t be half: it’ll be some more messy percentage, down to as many decimal points as you can be bothered with.
So what does all this have to do with journalism?
Well, to the extent that journalism is online, it makes its consumers subject to the same kind of analysis. For instance, Mashable tests alternative headlines for its blog posts using exactly the same technique that’s working for Adobe’s marketing department.
Then there’s the ability to understand and respond to precisely what audiences are interested in. As Hammell said, so far “very few businesses have captured the predictive power of social media.”
Techniques that Hammell said are being developed in the UK - to improve sporting performance through analysis of previous performances, for instance, or to decide about the siting of a new shopping centre using social data - are part of a new science that can be applied across many different fields. We can just know a lot more about people and how to reach them.
Of course it can go too far. Hammell described a retailer which got too clever by putting together the purchase patterns, ages and locations of its customers to produce an uncannily accurate “pregnancy prediction score”. That was used to pick out individual women who the retailer believed were likely to be having a baby or trying to get pregnant. Sending them special promotions produced complaints which caused “a total loss of faith in the business”.
It was Robinson who gave the most vivid picture of where all this is going. Mobile business has doubled for her agency in the past year, and things are moving so fast that she was talking about how much more data smartphones can now offer advertisers than in the bad old days - by which she meant six months ago.
A well-organised advertiser can filter its smartphone campaign in any number of ways to reach the right people: by the language used on the phone, by the apps downloaded, by location, or whether the phone is connected by Wi-Fi or 3G.
It’s a complex picture. The information available depends on whether the ads are being placed by, say, an app company, which has more information about its users, or by an outside business using the smartphone network more like a broadcast network, where there would be more limited ability to target.
But, however you look at it, as Robinson said, mobile data is “where a lot of the juicy stuff is”. And, given that more than half the UK population now has a smartphone, and about three-quarters of the ad-friendly 18 to 29 age group, the incentives to advertisers to make it work are enormous.
When Facebook issued its pre-IPO prospectus, it warned that advertising on mobile was an unproven business model (which makes it seem a bit unfair that its shareholders are now apparently exasperated because advertising on mobile isn’t an instant cash cow).
But Robinson’s presentation seemed to suggest a new way that mobile ads could come into their own by reaching the parts that other ads cannot reach. And she wasn’t worried about the small screen.
Instead, it seems that the intimate connection between a smartphone and its user (how often do you let someone else use your phone?) together with the use of big data to target ads, not just demographically but by time and location too, could be the key to the commercialisation of mobile.
For content providers - whether journalists, film-makers, musicians, games designers or comedians - if there’s a way of delivering ads that works well on mobile, there’ll be more demand for stuff to draw people to those screens.
So, even if you don’t particularly fancy seeing more ads when you look at your phone, it could be the sign of a new revenue stream - some of which might, one day, find its way into your pocket (next to your phone).