Interesting to read in The Guardian that there have been dips in the usage of social networking sites such as Facebook, MySpace and Bebo which echos anecdotal evidence from talking to friends, colleagues and clients.
Reading further, the article goes on to raise the issue about discrepancies between the various sources of website audience data:
“Nielsen Online only measures website traffic based on a panel of UK users at home and work - it does not cover usage in schools, universities and internet cafes, meaning that younger internet users are under-reported.
Other online traffic research firms, such as Comscore, do include users outside work and home, and as a result their unique user figures for individual websites tend to be significantly higher.”
I wonder if the mistake of equating a larger panel with higher audience figures has been made. This is the natural assumption to have made but one that is not necessarily correct since the point of market research is that you use a sample and then upweigh your data to the size of the total population - which of course will be the same in both cases.
However the Nielsen and Comscore panels do produce different figures, not because one panel is bigger than the other but that both panels are based on biased samples. Sample bias is an inevitable problem with using market research as a methodology – something illustrated by the different results of political opinion polls even when they are conducted around the same time.
This reminds me of a link to an article I saw on Katie Paine’s blog a couple of months ago which talked about these problems and how Neilsen and Comscore are having their methodologies audited.
My concern with this issue is that it misses a major point. At the end of the day when measuring an online article, I don’t care who views the whole website (clearly difficult to measure anyway!), I care who reads the article itself.
In measuring print media, most evaluation methodologies will make the assumption that they are both the same thing. In other words that everyone that reads a newspaper will read every article (sometimes qualified by “has the opportunity to read every article”!). This not far from a reasonable assumption to make – according to the last Quality of Readership Survey, the average person reads 79% of a newspaper’s pages.
But with online it is totally different. Let’s take The Guardian article as an example. According to Neilsen NetRatings 2,938,000 people view the Guardian online. According to ABCe (which is based on server logs rather than panel data and can therefore double count people who use more than one computer and include people living outside the UK) the figure is 15,955,000! This is further complicated by the fact that these are monthly figures and according to ABCe on average 789,000 view the site every day.
So which figure should we use. Well, none of these because the real number of people that have read this article is going to be way less. According to Alexa, the average visitor views just 2.6 pages of the Guardian’s website.
So how many people viewed this page. Who knows – well maybe the Guardian’s own IT department and possibly Google!
Until better data becomes available that more closely matches our consumption of this media, online measurement will remain very flawed.
Subscribe
Subscribe by Email