'Conscious commissioning': what The Times learned from deep analysis of its journalism

Credit: Photo by Jeremy Bishop on Unsplash

The Times has reduced its output of generic stories after a long-term, deep analytics project confirmed the obvious: readers value exclusives while barely tweaked press releases fail to engage audiences.

Taneth Evans, head of audience at The Times, spoke about the Content Review project at Hacks/Hackers event in London yesterday (22 January 2020). The research into different sections of the website was carried out in collaboration with Swedish analytics company Kit over 10 weeks.

The team examined what stories editors are commissioning and how readers engage with them. A cross-sample of 1,000 articles per section (including news, features, business, sport and travel) published over a 17-month timeframe was categorised into 12 'taxonomies' (or classifications) to understand the snap decisions editors make.

Across the board, generic stories failed to register much response with readers; embargoed stories without expert takes and court reports under-performed.

Before the research, stories commissioned for print would still get published online. That has changed, and now 20 per cent of these stories have been cut from the digital editions.

"This data backs up what we want to hear," explained Evans. "It’s telling us to invest in journalism, do more investigations and get more exclusives, don’t run wire copy."

At @thetimes & @thesundaytimes they categorised every piece of content they made with these 12 classifiers. A team of 8 people did this work. It took them 10 weeks. "It was not easy," says @TanethAutumn #HHLdn pic.twitter.com/21Q7XKsZ76

— Reuters Institute (@risj_oxford) January 22, 2020

The findings became more fleshed out after desk heads and commissioning editors from different sections compared their stories.

All of this data was based largely on dwell time index: a prediction model for how long it should take a reader to reach a set point in a given article. Other metrics looked at saved articles and comments.

"[The data was saying]: if it is not unique in some way, don’t do it, cut it," Evans said. "And then I got to the business section."

Although exclusivity was important - those big breaking stories over-performed content everyone else had - non-exclusive articles performed just fine when the team looked at subscriber reads.

"That suggests while we aren’t anyone's first source of news, we might be some people's first source of business news in the day."

It is something for commissioning editors to think about. In the business section, the data revealed that almost anything with data visualisation excelled, so now more business reporters are being trained to incorporate visual storytelling into their articles.

Writing style also matters as it reveals what content resonates with readers. In the travel section, surprisingly, reviews fared worse than 'escape' stories. Across sections, "what the f*** stories" (unexpected or freak stories) over-performed.

The data has changed the way editors work. On the foreign desk, for example, commissioning editors are resisting asking reporters for quick turnaround pieces, and instead, prioritising exclusives and original reporting.

Evans said she intended to relaunch the project with a view to filter the data by specific topics rather than sections. This could reveal what makes a science-based article more engaging to a reader: what tone does it need to strike and what type of headline works best?

"That will allow me to get all the science writers across both newspapers (The Times and The Sunday Times) - as well as those on news, features, business and sports - together and discuss it with the people who cover it day in, day out.

"That discussion will lead to different types of pitches, writing and commissioning."

The whole exercise was "resource-heavy", "a leap of faith" and was "not without push-back from executives". But editors generally welcomed insights the project has offered. Evans hopes that, in time, a model could be created which can automatically tag categories and topics. But factors like tone still require a human assessment.

"Watch this space, hopefully, we'll figure this out," she concluded.

Want to receive journalism news and job updates straight to your phone? Subscribe to Journalism.co.uk on Telegram on our jobs channel for latest job opportunities, and our news channel for a weekly digest every Monday morning.

Free daily newsletter

If you like our news and feature articles, you can sign up to receive our free daily (Mon-Fri) email newsletter (mobile friendly).