What I’ll describe in this post is a research project I’ve been meaning to tackle for some time, but the events of the past week have given it added relevance – for those of you outside Australia, I’m referring to the announcement of some 1900 job cuts at one of the country’s leading news organisations, Fairfax, and the hints at similar job cuts to come at News Ltd. In the wake of these announcements, there’s been plenty of discussion about the further decline of newspapers, and the shift towards an all-digital future, and some of my colleagues have articulated their suspicion that newspaper circulation figures in Australia have long been overestimated. Which leads me to wonder: is there a way to generate reliable circulation-style figures for Australian news Websites, by tracking the links being shared on Twitter?
We’ve been tracking tweets containing links to a handful of major Australian news sites for some time now; notably, this includes tweets where the actual URL contained in a tweet has been shortened to a t.co, bit.ly, or other short link, but eventually points to abc.net.au or one of the other sites we’re tracking. Using such data, we were able to document the impact of The Australian’s (partial) paywall on the circulation of its content on Twitter, for example. Recently, I’ve substantially increased the list of sites we’re tracking in this way, and we’re now covering almost all major national and regional news sites in Australia, as well as a number of opinion sites. In the following post, I’m presenting the first full week of circulation data, from noon on Friday 15 June to noon on Friday 22 June (we’ll move to a standard Monday to Sunday schedule soon, but I needed to get this analysis done over the weekend).
We’re now tracking tweeted links to a total of 26 domains: abc.net.au (from which we remove any links that aren’t directly news-related, such as abc.net.au/triplej; we also generate a separate figure for links toabc.net.au/unleashed), adelaidenow.com.au, afr.com, au.news.yahoo.com (which we split into the general news site and au.news.yahoo.com/thewest, the Web presence of the West Australian), brisbanetimes.com.au, canberratimes.com.au, couriermail.com.au, crikey.com.au, dailytelegraph.com.au, heraldsun.com.au, nationaltimes.com.au, newmatilda.com, news.com.au, ninemsn.com.au (also removing links which aren’t news-related, but retaining World Wide of Sport, A Current Affair, and similar TV shows), nit.com.au, ntnews.com.au, onlineopinion.com.au, perthnow.com.au, sbs.com.au (from which we similarly remove links that aren’t news-related, though I’ve left in links to SBS’s The World Game site, given that Euro 2012 is a newsworthy event), skynews.com.au, smh.com.au, theage.com.au, theaustralian.com.au, themercury.com.au, thepunch.com.au, and watoday.com.au. (If you think there are any major Australian news or opinion sites missing from this, please let me know. We’ve already added The Global Mail to our tracker, but too late for this week – and we’re staying away from regional and local news sites for now.)
I also need to point out again that our data gathering methods don’t allow us to track ‘button’ retweets – only ‘manual’ retweets (e.g. ‘RT @user …’) are included in our dataset. For that reason, I won’t focus particularly strongly on raw numbers of tweets here – but I’d suggest that the distribution of attention across the sites we’ve tracked is nonetheless a reasonable indicator of overall Twitter circulation (there’s no reason to believe that any one site would attract disproportionately many manual retweets).
I will say, though, that we captured some 150,000 tweets over this past week – and here’s how they’re distributed across the major news sites (not including opinion-only sites):
In a week of bad news, that’s a very strong result for the two major Fairfax sites of the Sydney Morning Herald and The Age – but it’s important to keep in mind precisely that this was a week of bad news for Fairfax, which may well have artificially boosted the number of links to those sites. It will be interesting to see whether this dominance continues into the coming weeks. The marketshare of News Ltd. sites is also interesting: individually, these sites don’t command positions as prominent as those of the two leading Fairfax papers, but in combination, the four top News Ltd. sites (news.com.au, The Australian, Herald Sun, Daily Telegraph) still account for almost exactly 25% of the news links shared. Finally, I’m also somewhat surprised that Fairfax’s online-only news site Brisbane Times outperformed its local News Ltd. rival, the print-and-online Courier-Mail, by some 900 tweets – again, it remains to be seen how much of this is due to the news about Fairfax itself this past week.
Let’s also have a look at the attention share of opinion sites, then. Here, it’s important to note that many mainstream news sites also contain opinion articles and columns, of course – but in addition to these sites, the last few years have seen a significant growth in dedicated spaces for opinion and commentary in Australia. Here’s how they compare:
Crikey rules the roost, clearly, followed by the ABC’s The Drum (or at least that majority of its articles by non-ABC commentators which are published under the abc.net.au/unleashed URL path; articles by ABC commentators sit in different URLs paths which we cannot easily distinguish from other ABC news content). The News Ltd. and Fairfax opinion sites, The Punch and National Times, are roughly even, while long-struggling independent opinion site New Matilda puts in a surprisingly strong performance. (Next week, we’ll have enough data to see how The Global Mail slots into this group.)
Update: I initially forgot to include The Conversation in the graph above, as some commenters helpfully pointed out; I’ve now updated the graph with a new version. And what a difference this makes: The Conversation had a very good week, clearly, especially around its coverage of the Fairfax upheavals – a very significant proportion of tweeted links to the site were to its articles about the Fairfax job cuts. Crikey and the ABC’s The Drum follow (or at least that majority of its articles by non-ABC commentators which are published under the abc.net.au/unleashed URL path; articles by ABC commentators sit in different URLs paths which we cannot easily distinguish from other ABC news content). The News Ltd. and Fairfax opinion sites, The Punch and National Times, are roughly even, while long-struggling independent opinion site New Matilda puts in a surprisingly strong performance. (Next week, we’ll have enough data to see how The Global Mail slots into this group.)
And here’s how the top sites’ Twitter marketshare looked on a day-by-day basis. Keep in mind that the two Fridays (15 and 22 June) each represent data only for half a day, so it’s no surprise that they’re considerably lower than the other weekdays. We’ll fix this as we move to a Monday-to-Sunday timeframe for future updates. What is notable from the graph, too, is an obvious drop in activity on the weekend (16-17 June), while Monday seems an especially strong day for sharing news on Twitter – but then, this was also the day of the Fairfax job cuts announcement, so we’ll see if that pattern holds in future weeks.
Update: I’ve also added The Conversation to the following graph, where a sharp spike in shared links on Monday and Tuesday is clearly apparent. Guess it helps having a former Prime Minister contribute an opinion piece about the Fairfax upheavals! Other than that, interestingly, for opinion sites the end of the week seems to be more important: Crikey peaks on Thursday (a good 100 of those tweets are about Julian Assange’s flight to the Ecuadorian embassy), along with The Punch, and The Drum (i.e. abc.net.au/unleashed, where a piece by Greens Senator Scott Ludlam on Assange was cited in 110 tweets) peaks on Friday.
So much, then, for this first edition of what I’m hoping to turn into a regular feature on this site. Of course, as we see a picture emerge from these posts of the Twitter link circulation marketshare for major Australian news sites, it will also be worth reflecting on what these figures mean; I’ll leave most of this for a later post. Clearly, though, we can’t assume that these observations translate straightforwardly into an indication of hits on these sites – the Twitter userbase is too unrepresentative of the wider Australian population for this, and what is shared on Twitter isn’t necessarily an indication of what is read (users may be more inclined to share highbrow than lowbrow content, for example, as they maintain a professional persona on Twitter). To get a full sense of the impact of these shared links, we would also need to take into account the number of followers each sharing user has, of course: a link shared by a users with a few dozen followers will necessarily have less impact than one shared by a leading account in the Australian Twittersphere. We’re beginning to gather some good data on this, and might bring follower numbers to bear on this analysis in the future.