The shocking truth about audience data: it will surprise you

Ophan’s screen showing every Google search (that we know of) driving people to the Guardian website.

In today’s Observer, Peter Preston is on great form, corralling the testimony of Paul Mason, Frederic Filioux, John Simpson and our own Martin Belam in pursuit of his key point: that the news industry is engaged in a clickbait-chasing, analytics-induced race to the bottom.

There’s a lot here to agree with and a lot to worry about. But I wanted to broach the specific point around analytics, one that ties in to Martin’s excellent piece on the difficulty of getting people to read about Lahore.

There’s a common thread in thinking around analytics and the newsroom which runs like this:

i) people like reading shit on the internet

ii) audience data will tell us that

iii) therefore any response to audience data must enforce a degrading cycle

iv) that’s just fucking life

It’s the lowest common denominator argument and it’s really attractive because it has an artificial and superficial nobility. Its us-against-them quality feels robust and solid in an industry where everything feels in flux and in immediate danger. It’s a certainty. A depressing one. But a certainty.

But it’s nonsense. And the really dangerous part of it is that it absolves us of our own responsibility for what our readers read.

The truth lacks the argument’s formidable certainty. It’s nuanced. It identifies the audience not as a single lump but as individuals, as varied as the world’s population, coming to us through a variety of radically different platforms. It takes effort. It recognises our responsibility to understand that audience. It recognises that the worth of a story can’t always be determined by its broad topic. And it demands that we look at the data.

Let’s take the idea of writing about Kim Kardashian, usually the first name to come up in any discussion about readership and quality. Let’s ignore the fact that there are many different pieces to write about her as an individual or as a symbol, and that even then the quality of the piece can only be determined in the writing of it. Following the lowest common denominator argument any piece about her will — must — find a huge digital audience.

But reach is unpredictable, dependent on context and subject to a huge range of factors. The assumption that Kardashian’s name will guarantee vast search traffic, for example, is faulty: if you’re fighting for that keyword you’re up against publications with far greater search authority and far more resource to allocate to that topic. If you’re a more traditional news organisation your chances of success are low. Even just knowing this one fact — something easily dismissed as a grubby and irrelevant piece of SEO strategy — changes the way we think about the topic, the potential audience and most importantly our editorial decision.

There’s another very straightforward question at the heart of all of this: does the greater ‘success’ of one piece of journalism inherently sully the success of another piece of journalism? On Friday this piece on Google’s ill-fated April Fool resulted in well over 1m page views. Yesterday Simon Hattenstone’s heartbreaking interview with the parents of Connor Sparrowhawk did about a tenth of that. Even if we didn’t also consider the time people are spending with a piece and the broader reaction to it, it’s clear that a direct comparison between the two can’t lead to any meaningful judgments or insights.

If we lament top 20s dominated by topics that don’t form the core of our editorial aims, then we have a wide range of potential responses in terms of what data you think is important, how you present it in your organisation and how you talk about it. But most importantly, just stop writing about the things you don’t want to cover. Be clear about what your aims are. Don’t ignore the data in a nihilistic funk. Don’t mindlessly let the data lead you to places you don’t want to go. We’re editors, not algorithms.

The Hunting of the God Metric: a five-minute talk from Newsgeist 2015

One of my favourite screens in Ophan is the live Google search results page. You can see it at the top of this piece. It pulls in every single search term (that we know of) that drives any referral to our website. To some this looks like a confirmation of public ignorance. But to me it’s a glorious reminder that our audience is made up of people from around the world with vastly different levels of interest in and understanding of specific topics. One of Google’s great strengths is that it’s nonjudgmental. As a former teacher I know the value of being able to ask ‘stupid’ questions. One of the biggest search terms around the UK budget this year was “what is the budget?” Looking at this kind of data isn’t about pandering to vulgarity. It’s an opportunity to respond to gaps in the knowledge of our readers.

The specific point around reader interest in the attacks in Lahore is an illuminating one. Martin notes in his piece:

“I was struck by the fact that despite leading the site with several stories about the Lahore attacks, it was not our most read story. In fact, Lahore didn’t even make the top five most read.”

To me, there’s encouragement here. Despite the fact that the piece was not the most-read on our site (at that very specific point some hours after the attack), our editors were leading on it. That’s an example of judiciously ignoring the simplistic message of the data and responding with nuance. We believe this is important; we can see that other things are doing well further down the page as they are; we are using this particular channel to particularly promote this topic. My next thought? Is there something we’re missing here to help readers connect with this story? Is it simply a question of distance or prejudice? Have we failed to promote or describe it properly?

There’s one thing I’m certain of. Making the effort to look at the data we have at our disposal is a better way of improving our journalism than making blithe and negative assumptions about what people want to read.