Skip to content

What big data means for media

May 22, 2011

Big data is changing the ways companies do business. So how will it affect the media?

THE COLOURS: A data visualisation of Wikipedia editing from IBM via Wikipedia

Two ways:

1. Journalism – as more datasets are released and software to query and visualise it becomes more common, data analysis will drive more and more real journalism. We’ll see:

  • More data spawning investigative articles
  • More data providing context to articles (eg crimes in different areas)
  • More data as standalone journalism. A lot of times the reader doesn’t actually need a story to provide context, and the data visualization does a better job by itself. Let the reader and their social media buddies decide what’s interesting instead of a reporter.

2. Business models – Data is following its own version of Moore’s Law right now. Collecting, storing and analysing information is becoming cheaper and faster, and at an accelerating pace. That means it’s going to get easier and more economical to track more of what you read/watch/hear/do online. It’s also going to get easier for companies to link up their independent databases to gain new insights into consumer behaviour, and exteremely targetted advertising.

The data should be anonymised – in both meanings of the word “should” – but you’re not being paranoid if it makes you uneasy.

Datafication will also accelerate the changes encouraged by the internet, where page views and other metrics were brought down from abstractions to provide real insight (and KPIs) to ordinary journalists. Before the internet, editorial judgement and occassional reader surveys were the way to gauge the popularity of different content. Now you can see what rates and what doesn’t in real time. There are upsides and downsides to this, as we’ve all read about.

But there’s a more fundamental change on the horizon. Being able to track everything each reader does on your site – and maybe some info on what they do on others – is another nail in the coffin of traditional editorial judgement.

In the bad old days it was the editors’ jobs to figure out what went on what newspaper page. Nowadays it’s web editors deciding what should lead the site or a section homepage. But when everything is tracked, the data does that for us. Web editing will increasingly become low-skill content loading, which will quickly be replaced by technology. I’m a programming noob but know enough to write an algorithm which ranks stories based on their recent clicks.

Within 10 years I expect online newsrooms to start whittling down their web editing staff and putting more resources to reporting – exactly how it should be. There should be as few staff as possible between a reporter and their audience. Web editors will be there to massage the system and give it personality, a veneer of humanity over a machine that handles all the grunt work.

Another point: News will soon enough be able to be completely personalised. Each person has a different set of stories served to them. Soon enough, editorial judgement for a mass audience won’t be a very marketable skill anymore.

So what’s a web editor to do? Knowing how to load stories and extra content won’t be enough to keep you in a job. You have to learn skills that technology won’t be able to easily duplicate in the medium-term, like growing commmunity and encouraging discussions that keep readers on site.

Also expect the commodification of news websites serving similar audiences. As designers understand what makes people click, sites will increasingly optimise their designs to extract every last click, every last second of engagement from users. The same way most newspapers fit into a handful of design categories, news websites will too soon enough. Even more so, because the traditional editor’s discretion (“I like that and not this.”) won’t count for squat. The data will prove what works best.

Right now the media world has hundreds of different content management systems to handle its content, because there are so many different ways media sites want to show that info. But as design starts to homogenise, these CMSes will get whittled down to a handful of major players who can best (usually most cheaply) serve up content into those popular templates. This will slow down news design innovation online until things are broken up by the next revolution.

And once the design and CMS approach have been cemented into only a few real choices, you’ll be able to buy your major news website out of the box as a managed service. tweaking the colours to suit your brand.

The structure of individual news articles will also change to optimise readers and revenues. The technology is already here to serve up two different versions of a page to see what rates best, it’s just not economical really yet to do it for specific articles. But as data costs drop, it will become common for companies to experiment with different article forms, and decide which forms are most efficient (lowest cost/revenue) for each subject.

Subjects too will become increasingly tailored to revenue rather than raw audience size, as they are now. For example, covering rugby in New Zealand is guaranteed to bring in lots of readers, but is it worth taking one of your several reporters away to cover ice hockey once a week? That’s easy to find out if you know the average output of each reporter and ad returns of each topic. We can do this now, but it’s usually not economic. In the future though, the numbers will decide what topics journalists cover, and how much they are covered.

That might paint a bleak picture of the future of journalism for you, but dataification is not a bad thing. Computers aren’t going to replace human beings except in areas where they should replace them. Computers are good at repetitive tasks, and not at all good and understanding how people think and behave, or understanding what a story means.

And that’s the kind of skills online journalists in the future will need – more psychology than programming (though as much as possible you should know both). Find a field and be expert enough to provide context, and your job will be safe.

All the grunt work, that can and should and will be outsourced to machines.

No comments yet

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: