"You'll see. Time spent on sites will settle the issue of audience measurement. That is where everyone will ultimately come to an agreement". Thus spake the head of Mediametrie /NetRatings in France. Last week, we were having another passionate discussion about the way we measure websites audiences. My view: the always changing methodology, the discrepancies between site centric measurement (software at the server’s root that tracks clicks) and the user centric measure (a polling system such as Nielsen’s), all contribute to the confusion -- and to the resulting deflation of advertising prices. His point: there is no perfect system, we’re in the throes of a continuing evolution, and we are working relentlessly to improve our measurement methods. Well. I’m happy to give credit to the gent -- his name is Bruno Chetaille, I’m ready to believe he’s genuinely committed to providing the best websites measurement tools possible. He's smart and open-minded, and his idea of time spent on the site as the ultimate gauge triggered my curiosity in two directions:
- Which sites generate the best "stickiness", i.e. users staying for the longest time ?
- More important for this note’s purpose, how does the news segment behave in this regard?
To get a closer look, I massaged recent French market Nielsen statistics for November 2008 (latest figures available). I believe most of the trends and underlying audience engines do apply to other markets as well.
1. Starting with the background: most viewed categories as measured by Unique Visitors (UV) per month. Chart n°1 below shows three trends. The biggest hits are search engines, portals and email services collecting about 85 m UVs on November. Nothing unexpected here. Then came typical Internet services such as MSN, ISP and mobiles operators and related apllications (personal home pages, webmail...)
News land in a distant 14th place with an audience of only 7% of the combined categories ahead; with less than 20m UVs, news is also outrun by the "Blogs and Community sites" category (altogether 23m UVs) that includes Facebook (9.5m UVs).
Two conclusions emerge: 1) News is not a major audience engine in itself and, 2) News definitely must channel some of its content to blogs and community platforms, whether through its direct operations (online newspaper X or Y owning a big, top class, network of blogs, or through the use of social networks and community sites to channel traffic back to itself)
2. How long do people stay, and on what? The chart below takes in account the amount of time spent on domains, per visitor, in November.
This is just a selection of sites: their ranking in time spent appears on the yellow background. Obviously, the two big winners based on the hours spent each month are the online game King.com (almost 5 hrs) and Facebook (two and a half). Then, many classifieds such as Leboncoin.fr (a free classifieds site owned by Schibsted) or anpe.fr (the publicly operated job site) are doing well, but they are down to roughly 30minutes of monthly connection time. The video sharing sites are doing fairly well: YouTube ranks #48 in time spent with 31 minutes for the month and its French competitor DailyMotion gets 16 minutes (we could have expected more for those two, but the system for embedding videos that is available on any blog could explain the lower numbers).
How are news sites doing by this metric? First, let aside the web of L'Equipe, the main French sports daily: sport fans is an audience in itself, and each user spent almost 40 minutes a month on it (it ranks #38 in time).
In the general news sites segments, the first one lands only at the 102nd place: Lemonde.fr, the eponymous site of the big French paper is the second newspaper site in France (4.5m VU) but people spent only 15 minutes on it, two more minutes than on Le Figaro, number one in audience (5.2m VU). Further down in the chart, after classifieds and financial information sites, we find the third and fourth newspapers: respectively 20 minutes, the most read paper in France, and Liberation (about 11 minutes per month for each). The two most active pure players Rue89.com and LePost are below 10 minutes of monthly viewership.
Keep in mind that these are average numbers. It is quite likely that we are in a 30/70 configuration making the average a little misleading. Here I mean a small fraction of the audience, of the UVs, 30%, spending most of the time on news sites and 70% spending little time. Something like the proverbial consultant who looked at the river’s average depth, 30 cm, and went on to drown as he attempted a crossing. Still....
(A description of each of the sites is at the bootom of this page )
3. Finally, an even closer range picture: how long users spend on each page? It yields interesting findings.
- Numbers are small. In the news category, any given page captures attention for no more than one minute. Again, these are averages, and when people see an interesting piece of long text, many of them print and zap the page.
- The richer the content is, the more people spend time on the page.
- But rich content, considered time wise, come under various forms: for instance, two pure players are doing well. One (rue89.com) is offering rather long and focused articles targeted to a politicized audience. The other (LePost.fr) features a large volume of comments and user-generated content -- and something like four journalists.
Have a look at the chart below for details.
Let me wrap up with a couple of considerations. These measurements are to be handled with caution. A large number of factors can affect the time anyone will spend on a site. A news site can suddenly jump in time spent because it hosts a successful game. Others have smaller pages: it can take several pages to read one article that could be displayed in a single-page format on another site (therefore increasing the time spent per page). One interesting measure could be print-format data since many people are actually printing web contents.
This shows how much work remains before we get reliable measurements on the Internet. Obviously, we can't bundle a YouTube video, a Washington Post 2000 words profile, a gossip The Drudge Report, or a party on an online game site in the same number. But for media, it is more urgent than ever to find reliable metrics, most likely a mixture of site and user centric measurements. This ought to be a priority for publishers. —FF
Description of the selected sites