The Fairfax New Zealand news Web site Stuff hit a record milestone of more than 2 million visitors in March 2016. This solidified our position as “New Zealand's leading digital brand.” We found ourselves in the unique position of not only being the largest domestic Web site, but only falling behind the two major contenders, Google and Facebook, in terms of monthly unique audience.

Enviable though this position is, we are not impervious to the challenges publishers face today with the rise of social platforms and the shift to mobile. Rather than remain stagnant, we are constantly on the search for strategies to sustain our growth as well as drive higher levels of content consumption and engagement.

Over the past year, we spent considerable time analysing the article-level data in our content authoring platform, as well as audience behaviour data in our analytics tool, to identify trends that could drive strategic decisions for both the editorial and product teams.

Strategising publish times around peak user traffic times was a major goal from data analysis.
Strategising publish times around peak user traffic times was a major goal from data analysis.

We started by looking at where Stuff’s audience comes from, and how its composition has changed over time. Our desktop audience is stagnating, while the mobile audience is growing with a significant increase of traffic coming via search (predominantly Google) and social (predominantly Facebook).

However, these proportions were still not as high as industry standards. We realised we needed to spend more time tracking our social posts and analysing our social audience to grow this number.

This finding also led to the decision for an 80/20 split between resources devoted to the mobile and desktop sites, with the aim of converting “rented” audiences from search and social to Stuff readers. We looked at what users from those channels read and watched, as well as studied their user paths, to understand how we could increase reticulation on the site through changes the content we presented to them.

Another exciting revelation was the disproval of the myth that celebrity news contributes to a greater percentage of our content consumption than major news events. We found that not only did articles related to major news events generate more page views (on the whole and on a per article basis), but also the longevity of interest in these articles far surpassed that for celebrity news.

In fact, between January and November of 2016, only four of the top-15 news events were about celebrities, with hard news showing long-term appeal versus the fleeting appeal of celebrity news. In particular, our live blogs perform exceptionally well on days of breaking news events, with a high proportion of visitors returning to the blog one or more times during the day. 

Contrary to general belief, Stuff found that celebrity news stories received far less engagement than breaking news.
Contrary to general belief, Stuff found that celebrity news stories received far less engagement than breaking news.

Another interesting finding was that during months when we didn’t have a significant breaking news event, our highest traffic days were those on which the top-10 articles covered disparate topics. These articles brought in unique audiences with minimal overlap, so only a small percentage of users went on to read all of the top five stories. This argued the case for breadth in content and a well-curated content mix.

This inspired us to dig deeper into how Stuff’s content mix was influencing the reader experience. We wanted to know that we were devoting resources to create content that gave us the optimum return on investment in terms of audience and consumption.

To perform a thorough content audit, we created a dynamic Qlikview tool to merge authoring and audience behaviour data that would reveal how our newsroom, vertical, syndicated, and contributor content was performing. This gave editorial insight into quick wins for improving our ROI.

For instance, we found that a large proportion of low-performing articles only accounted for a small percentage of page views. Head digital editors conducted workshops with individual newsrooms and verticals, covering the topics of low-performing and high-performing stories, optimising publication day and times, as well as training on homing, tagging, and promotion of stories.

It was also noted that our biggest audiences occurs outside of office hours (mostly mobile), but we had a lack of content being published in the evening. A heat map was created to identify the times of day when specific content performed best, so we could tailor our output around that. Our editorial team made a commitment to boost evening and weekend content and to tailor publishing times around audience interest.

As valuable as these findings have been, they have also revealed further questions and a need for additional insights that can be actionable in real-time. Creating a tool that allows for real-time insights will be our primary focus for the rest of the year, and a pivotal step towards incorporating data into our day-to-day strategic decision making process.