This is the decade of data for the news business. We now have the tools and in-house knowledge finally to draw robust and meaningful insights from the deluge of data available to us.
How many publishers can honestly say they are making the most of it?
My observation is that data is growing in acceptance but still has a long way until it is driving editorial and business outcomes to the extent that it should. It is also newsrooms, the heart of content businesses, that are the obvious laggards.
Many, incredibly, still seem to think that gut instinct always trumps data and that to look at the analytics somehow negates their skills as journalists. To which I would say, look around you: the world’s dominant companies are data companies.
Amazon, Google and Facebook didn’t get where they are today by following gut instinct rather than data. Apple knows its data is so powerful that it has made a selling point of not using it for ill.
Getting away from generalities though, how can you practically increase the usage of data and insights in your newsrooms?
1 Remind the newsroom that data is your readers
In the halcyon days of print, there were only three data points: print circulation, ad revenue, and whether the editor liked the story. For most journalists the latter was the most important. And if they kept him or her happy, then life was good: your journalism was measurably of a high standard.
Of course, there was the small matter of letters to the editor – in truth a tiny number compared with the number of readers – and calls to the newsdesk from angry readers. If perhaps five people rang in, phones were judged to be “ringing off the hook” (I realise that may be a peculiarly British expression for “a lot of calls”).
I remember an editor once telling me that loads of readers had complained because they couldn’t find her section on the website that I was running. There were “a huge number” of complaints, I was told. “Okay, why don’t you record them all and we’ll have a look in a month or two,” I said. The eventual spreadsheet that was handed over had five names on it, two of them members of the editor’s team.
We were operating in a world effectively without data. Yet now we have almost too many data points for non-specialists to comprehend, much of it seemingly arcane to journalists, who tend to be generalists and are looking for quick analysis of a situation, not a deep dive into data.
How do we get them over this first hurdle?
I recommend heading straight to the most obvious fact about data, which is that it is your readers.
Nothing, not an editor’s hunch, nor a bulging postbag, nor an unusually busy call centre, will tell you more about what stories they engage in than your user data. For the first time in history we actually know, in pretty minute detail, exactly what our readers are doing with our copy, what they read, for how long and what they do after they’ve read a story.
If that’s not enough to get the start of engagement with data, then, well, it’s going to be a really uphill struggle.
2 Make sure the data is robust
Journalists are very good at taking a statement and picking holes in it. A trait for which I have only praise: it’s what gets them stories.
However, they can act the same with data. If there is the merest hint that the data is not telling the full story, they’ll be onto it in a second and, often as not, reject what you are trying to tell you.
At The Times, we had a period of perhaps a couple of years when the data science team was busy cleaning up our data. To cut a very long story very short, there were event tags in the code of our apps and websites that were firing inconsistently. We consequently had data that was not comparing apples with apples. But we still had to talk about it.
This was tricky. I guess we could have kept quiet about the issues we were having. But ultimately I think it helped us to show our working through this period: we told people that while they might not be able to rely on the individual data points, they could trust the trends, and that soon we would have gold-standard data. Because we had explained what was happening, when the data was finally robust we were able to talk about it with confidence and be believed.
3 Be clear that you are being data-informed not data-led
The primary fear of journalists and some editors is that by “following the data” they will be led into a world of clickbait. “They’ll just end up writing about Harry and Meghan.”
An editor once told me that this was why he was wary about giving data to the newsroom. My response was: “That’s why we have editors – you get to stop them writing clickbait.”
Nobody is suggesting that journalism should be purely led by the data. Instead it should be informed by the data.
If the numbers show that your readers like reading about a particular subject, what on earth is wrong with giving them more of it. You, as editors, decide how you report on that topic and how you present it, after all.
Sure, there is an argument about how one might get led away from journalism that might be deemed good for the public. But my experience of data is the opposite.
We always found at The Times that the kind of stories that performed well with our subscribers were those which had authority in the reporting, broke new ground or offered a distinctive take on events. In effect, the data described the reasons why most people in our newsroom got into journalism.
I accept that sites that are solely chasing page views might have a different experience, but in the end they too will be rewarded for the engagement they foster in their users, and you don’t get that from data-led clickbait.
4 Offer your teams insights alongside the data
A barrage of data without context is no good to anyone. I have seen so many dashboards in my time which offer many different data points, but where the net effect was not greater understanding but a form of data blindness.
I’m sure there is a psychological description of this but often when faced with too much of something people have a tendency to look away completely. The feeling is “there’s too much to get my head around”.
We countered this by concentrating on the insights that the data was delivering, whether that be by using indices, which provided easily understandable context for page views, dwell time, the number of comments and so on, or by simple colour coding (red, amber, green) for raw data.
The net effect was to take the “heat'' out of the data and concentrate on what you do next with the information it provides.
Don’t focus your data usage on what just happened, but on what happens next.
5 Put data where people work, not just on big screens
An array of big screens with data on them is often seen as a sign of digital virility by editors-in-chief and senior executives.
I remember one editor constantly asking: “Where are the screens? Why aren’t they up yet?” We had been taking our time to make sure the data was useful to the newsroom. He didn’t care what was on them, and never looked at data himself, but it was his way of proving to others that he had a digital newsroom (he didn’t).
Other newsrooms have seemed to specialise in putting data screens in places where they are practically invisible to their staff. I once visited a newsroom in New York and was amazed that a majority of their screens could not be seen by anyone doing any actual work. Instead they were effectively there to impress visitors.
This data window-dressing is pointless. Instead, put your data on the screens of the people who will use it.
At The Times, the main viewport for Inca, our award-winning analytics tool, was a humble Chrome extension, available on every workstation in the newsroom. It meant people could instantly refer to the performance of any given article.
They also never had to worry which dashboard this or that data was in, or remembering their username, or their password. It was just there.
As with most things in the world of digital news, data benefits from being kept simple.
That said, this is Data for News 101. We haven’t even got into audience-based commissioning. That’s for another day.