Newspapers suffering from bad math
John Duncan, the former managing editor of The Observer in Britain, is a newspaper consultant who writes a blog called The Inksniffer, and has lots of interesting things to say about what the industry should be doing. In one of his latest posts, however, I think he does more harm than good — or rather, he starts off on a good foot and then goes off the rails (to mix a metaphor). The post is entitled How Internet Metrics Promote the Myth of the Dying Newspaper, and his argument is that poor Internet traffic measurement makes the paper-Web balance look worse than it really is.
He is quite right about that, of course. As I have mentioned before (most recently in this post), Internet traffic rankings from all of the major firms — Compete, Alexa, comScore, Nielsen and Hitwise — are flawed in all sorts of ways, and combined with a continuing obsession with page views rather than unique visitors, it’s easy for Web audiences to look larger than they really are. And if John had stopped there I would have been totally onside. But after going through the numbers with admirable dedication, he arrives at the conclusion that “in the UK there were 310,788 people buying the Guardian or Observer on average each day in April 2007 [and] there were 270,576 reading guardian.co.uk online.” Not bad, right?
But then, he says that “We need to compare readers to readers, not readers to purchasers,” and falls back on what has become accepted wisdom in the newspaper business: that anywhere from three to five readers look at every purchased copy of a newspaper. In other words, the Guardian gets to multiply its readership by three at a minimum, which gives the paper more than 970,000 readers and online only 270,000.
As I mentioned in a comment on John’s post, this is complete rubbish — some of those copies might be read by two people, or maybe three, but plenty of them are read by no one. Steve Yelvington agrees with me, as he points out in a comment and on his own blog. He says: “the notion that a newspaper’s daily print sales figures should be multiplied by some factor to derive actual readers is wishful-thinking crap.” As Steve says, if we are to get anywhere, we need to be clear about what we’re talking about.
Related posts:
(On Jun 22nd, 2007 at 1:41 pm)
Hi Matthew: Well, you’re right, sort of. I’ve never been a fan of readership numbers either and my communications with the UK’s National Readership Survey only reinforced my confusion about how 2 or 3 people manage to read every copy of every newspaper. But it still isn’t enough to compare a “unique user” of a web-page with a purchaser of a newspaper and say they are comparable metrics.
I don’t believe the NRS readership figures are deliberately exaggerated. The NRS in the UK is a joint effort by advertisers and media owners, so I don’t see why it would do that. But I suspect that their definition of having read a newspaper may be something that allows anyone who saw a copy lying around the house or office yesterday and used it to check the football scores to tick a box marking them out as a reader.
I would say that the number of newspaper readers per copy is probably greater than one (I’d put it at about 2 maybe less); while the number of unique users who are actually unique (ie different people) is probably less than one (cookie deletion, repeat use at home and work etc). The exact numbers, frustratingly, are still elusive. Can’t wait for August though!
(On Jun 22nd, 2007 at 10:38 pm)
Thanks, John. For what it’s worth, I think you’re right too — sort of
I’m not sure that the number of newspaper readers per copy is greater than one — in many cases, I think it is less than one, but I’ll give you that point regardless.
I think the more important point is the one you allude to, which is that even “readers” of the paper don’t actually read most of what newspapers and advertisers convince themselves they do.
Online, by contrast, it’s quite easy to see what people actually have read — on an individual article basis — and how long they read it for (although in some cases, such data may measure people who kept a page open but didn’t actually read it).
I suppose we will have to wait for the new brain-implant technology to answer this one conclusively