Second Life figures cast doubt on Ofcom report
Virtual monitoring is virtually impossible
Disputed figures put out by broadcasting industry regulator Ofcom last week, in respect of Second Life usage, raise questions as to how seriously we should take Ofcom reports.
However, the debate goes beyond questions of mere accuracy, raising issues around how we can generate meaningful measures of social networking sites.
Last week, Ofcom gained mega-exposure and mucho publicity with their 2009 Communications Market Report (pdf), which they described as "a reference tool to track the development of the UK communications sector". They added: "the report also provides an important context for the work that Ofcom undertakes in furthering the interests of consumers and citizens in the markets we regulate".
It is therefore an important document: one that seeks to establish universally acknowledged yardsticks for online behaviour, as well as one that claims the right to inform the big debates about government policy in this area.
Here at Vulture Central, we were intrigued by the report’s claim that in the space of 12 months, UK user numbers over at Second Life had fallen by 67 per cent, whilst average usage of Second Life had also fallen by 68 per cent (from 28 hours per month, to 9). If true, this would represent a devastating blow to Second Life, bringing aggregate usage in the UK down to just over 10 per cent of what it was a year ago.
Second Life utterly reject these figures. According to a spokesman, user numbers and usage are both up. The latter rose over the year by just over seven per cent, from 35.5 hours per month to 38.6 hours per month.
They ought to know, as online usage of the SL app is recorded to the nth degree in their central database. Nielsen, who produced the figures for Ofcom, countered that it is a common misperception that third party research must be more inaccurate than in-house data. They suggested that their take on SL usage might be a better representation of reality than the raw SL figures.
For instance, where an individual remained logged in, but had other windows open, Nielsen would stop counting data from that individual after a given time interval.
This echoes controversy that dogged the monitoring of TV audience behaviour in the 80s and 90s in respect of advertising. Some methods were very good at identifying what TV programmes were being played at a given point in time – but research highlighted how what was playing on the TV was not necessarily what was watched, as individuals left the room during commercial breaks to make tea or go to the toilet.
Measurement of online activity initially followed the model adopted by the direct marketing industry, with the focus being very much on recordable events: page clicks, click thrus, time spent on page, etc. As social networking becomes a much more commonplace activity, bodies representing marketing professionals – such as the Institute of Direct Marketing – have been asking whether traditional measures can still be applied.
How is one to count game-playing on Facebook, where the game being played mostly just sits within a Facebook shell, whilst actually being hosted on another site altogether?
As for Second Life, how is one to count time spent listening to music – or even leaving a window open in order to "read back" later on? Nielsen argued that their measure of SL activity, which aggregated app usage and online site visits, was the "correct" way to go, since it produced a measure that was directly comparable to other social media.
However, SL responded that such a measure sacrifices meaning for consistency – and produces figures that are essentially meaningless, or just wrong.
Neither body were especially reassuring when challenged. Ofcom’s response was to pass the issue straight back to Nielsen: they may have produced the report, but if we wanted any further information on how Nielsen got its data, that was "a matter for Nielsen".
Nielsen’s explained in great detail why their approach was most likely "right" – without at any point in a very long exposition pausing to wonder what they were reporting – or what it meant.
In other words, they were prepared to defend to the death the "accuracy" of their method (open windows, active keystrokes, etc.) without ever pausing to wonder what they were measuring, or what the point of their measurements were. A little like asking how much a banana weighs - and receiving the answer: "yellow"."
Initial questions asked whether the report was "fit for purpose": that is, if it was so far out in respect of SL, what credence could be given to other figures in it. Their combined response suggests that the figures reported might not actually be "wrong": simply that they do not relate to any useful purpose. ®
Sponsored: Navigating the threat landscape