Newspapers on the web: Why daily metrics don't answer online measurement's biggest question.

Secondly, I'd like to see an official breakdown of the UK and foreign audiences released by the ABCe. And if they can manage it, I'd like to see them have a stab at incorporating a realistic "margin of error" to their numbers based on research into cookie deletion and dynamic IP addressing issues. The science behind ABCe is subject to errors that don't occur when you are measuring copies sold. It's incumbent on ABCe to reflect that in the figures they allow newspapers to use if they intend to offer an ABCe as a number you can compare to an ABC.

One step forward. Still a long way to go

Is internet advertising ready for its first "Newsday moment"?
Posted by Sniffer dog at 6/22/2007 10:44 AM and is filed under Advertising pricing,Display advertising,Market Research,Readers,NEWSPAPERS,Online newspapers,Advertising,Technology
If you want to highlight the double standards applied to audience measurement online and in print I'd like to compare the advertising industry reactions to Newsday's misstatement of their print circulation, exposed in 2004, to the revelation by ComScore that the counting methodology used by some rival internet metric companies to measure unique users may have been out by as much as a factor of 2.5. Nielsen NetRatings announced similar findings in 2005.

Two arrests, $90m refunds in one. Total silence in the other. You know which is which.

ComScore and Nielsen, the metrics firms who put out the (separate) research, have an obvious interest in the story. Both firms use a panel system to measure audience while other companies try to identify unique users via cookies/registration data and IP derived system. Both sides claim the other side's methods are flawed.

It's not entirely the same situation of course. In one the publisher knew it was cheating on an accepted metric. In the other, one form of the metric itself is being challenged. But in both cases an advertiser may claim to have bought advertising on the basis of publisher backed numbers, which subsequently turn out to be inflated.

It wouldn't necessarily be fraud either. You'd have to prove that someone in a newspaper knew that they were using figures that exaggerated their audience in order to get more money from advertisers. And then prove that advertisers relied on the numbers, and had no other way of knowing what the true numbers were. I think that would be a very difficult thing indeed to do. So the Newsday moment could be some time coming.

==============================

Anyway, let's get back to ComScore. They sound like they care about accurate metrics so maybe we should explore how their numbers stack up to the calculations we made for the Guardian yesterday.

It turns out that The Guardian are ComScore clients. Digital supremo of Guardian Media Group Simon Waldman's January piece contains a ComScore graph of "Percentage reach of UK monthly internet users" that suggests that the Guardian has about 9% reach among UK internet users.

But 9% of what? Well, ComScore's own monthly unique user numbers for the internet in the UK say there were 31,150,000 unique monthly users of the internet in the UK in April 2007. And 9% of 31.150m unique monthly users in the UK? 2.804m

That's odd. This number is substantially below the 5.7m UK users listed on the Guardian's ABCe-certified AdInfo for May.

It's made more puzzling by Simon's belief that
"we feel their [ComScore's] panel has both the right scale (around 29,000) and takes the most scrupulous approach to weighting (in other words to ensure that their data is representative of the audience as a whole, rather than simply aggregating ISP data as Hitwise does)."

Obviously the ABCe data is even more "scrupulous" ... and has the added benefit of doubling the ComScore measurement thrown in.

It's not the Guardian's fault. It's the lack of an agreed standard that is to blame. If you were offered several different measurements of your audience all of which could claim legitimacy and are provided by independent reputable companies, which one would you choose to use? The biggest or the smallest?

The ABC for print publications works because everyone in the UK publishing and ad industry agrees it's more or less accurate. And everyone gets measured the same way. Punishing cheaters is in everyone's interests because it challenges the advertiser's faith in the whole metric. Punishing cheaters in the online world is not really possible because no one agrees on what cheating even is. But these are still the numbers that whole media companies are being gambled on.

I'm usually a fan of the benefits of competition, but strangely the web metrics business could be the one area where competition is actually bad for the customer (the advertiser) because it destroys his ability to define truth. I, as an advertiser, can't tell if the Guardian's unique monthly usership in the UK is 2.1m (Content/Alexa), 2.1m (Nielsen), 2.8m (ComScore) or the 5.7m (ABCe) that appears on the paper's AdInfo website.

In the end, it probably doesn't matter that much, because ad agencies have access to their own data and may choose to take the newspaper's declarations with a pinch of salt. They may want to dice the audience a completely different way to unique monthly users. And they can do the same ComScore sum I just did and chuckle to themselves at the newspaper internet audience claims much like they chuckle at our ratecards.

But I'm not sure the people running newspapers or fighting for print resources against the internet land grab are as well-informed. They may not even be aware of how fragile the self-serving numbers being propagated by their online departments are.

The numbers, at the very least, certainly aren't strong enough to bet the company on. But that's what's happening all the same.

Comments

Add New

RSS