There will be another post on the continuing Queensland election campaign soon, but before we get to that, I wanted to push out another quick snippet of analysis which I’ve been hoping to get to for some time. For several months now, we’ve been capturing tweets which include links to a number of major Australian news and opinion sites. We’re able to capture such tweets even if the links are wrapped in one or more layers of URL shortening (via bit.ly, t.co, etc.) – so even if the tweet itself reads something like
Opposition gets Malaysia deal wrong http://t.co/DH3quv86
it will be included in our dataset.
(In fact, just to work through this example, here’s how that link resolves: http://t.co/DH3quv86 > http://bit.ly/omcBt0 > http://www.google.com/url?sa=X&q=http://www.theaustralian.com.au/national-affairs/opinion/opposition-gets-malaysia-deal-wrong/story-e6frgd0x-1226137247377&ct=ga&cad=CAcQARgAIAAoATAAOABAm_jC8wRIAlAAWABiBWVuLVVT&cd=GuZ8wAwkEaQ&usg=AFQjCNFJehe4mpkMiFiNCWt8Ey9kWT42jw&utm_source=twitterfeed&utm_medium=twitter > http://www.theaustralian.com.au/national-affairs/opinion/opposition-gets-malaysia-deal-wrong/story-e6frgd0x-1226137247377. That’s a lot of unshortening…)
I’m not yet able to do a great deal of comparative work across multiple Australian news sites, simply because (as you can see from the example above) unshortening Twitter links takes a substantial amount of time. However, there’s one particular question which I’ve been very curious about: do our data show any evidence of changed usage following The Australian’s decision to place some of its content behind a paywall? There’s been a lot of conjecture about whether such paywalls actually work – for other NewsCorp titles, they’ve been claimed to have failed (the company itself won’t release any real figures beyond initial sign-ups, of course). So, is there anything we can say about the Australian situation?
The Oz’s paywall was introduced on Monday 24 October 2011, and follows a ‘freemium’ model: the site offers free access to generic news, but restricts unpaid access to what The Australian describes as its premium content: “analysis, opinion and more specialist material”. In particular, it limits click-throughs from other platforms: the paywall allows “access up to five premium-content stories a day via Google and … one free click-through from social media site Facebook.” Twitter and LinkedIn users get no such limited free access at all, apparently.
This mixed-access model doesn’t make measuring the real impact of the paywall any easier. However, as a shortcut, we can examine the number of links to articles in the paper’s ‘Opinion’ section – the majority of which are paywalled. Links to such articles will usually include ‘/opinion’ somewhere in the URL – and so, with these caveats, here are some figures on the amount of such links being shared over the past few months. To adjust for day-to-day variations, I’m providing aggregate figures for each calendar week here:
24 October 2011, the day the paywall was introduced, marks the start of calendar week 43 – and we do indeed see a marked drop in links being shared. Even adjusting for seasonal variations as we enter the long Australian summer lull (and ignoring weeks 50/2011, when we suffered some server trouble, and 0/2012, which consists of only one day), the contrast is clear. During weeks 38-42 (19 Sep. to 23 Oct. 2011), our dataset contains an average of 975 ‘Opinion’ section links per week; during weeks 4-8 (23 Jan. to 26 Feb.), that average has shrunk to 486. Remember in all of this that some ‘Opinion’ links did remain freely available, of course – so a more comprehensive paywall would be likely to reduce numbers even more.
The picture becomes even clearer when we focus only on retweets. More so than through original tweets linking to a given article, it is through retweets of those tweets that the article is circulated widely, and gains greater exposure and (potentially) impact – but if the link in the original tweet is paywalled, who (other than fellow paying subscribers to the Website) are likely to retweet it? Not many, as it turns out:
The average number of retweets linking to ‘Opinion’ pieces changes from around 368 per week in the first five weeks of our timeframe to a meagre 90 in the last five weeks, with week 43 showing a steep drop. The usual disclaimers apply here: our method of capturing tweets using yourTwapperkeeper will only capture manual retweets (and I’ve filtered for tweets in the format “RT @username” to generate the graph above); we have no data about ‘button’ retweets, but it seems highly likely that the drop-off pattern will be very similar.
This, to me, shows the impact of the paywall most obviously. Those with access to the full articles might still tweet about them (and presumably, by extension, share them on Facebook and elsewhere) – but if their social media friends and followers can no longer access those articles, they’re also no longer going to re-share those links to their own networks of connections. By paywalling its opinion pieces, The Australian is effectively removing them from wider circulation. If the intention of the paywall is to shut out the users of Twitter and other social media platforms – mission accomplished. But if by doing so The Australian aims to encourage those users to sign up for premium access themselves, I doubt that’s working…