Webster's Dictionary of Audience Exaggeration: How internet metrics promote the myth of the dying newspaper.
Posted by Sniffer dog at 6/20/2007 3:01 PM and is filed under Advertising,Trends,NEWSPAPERS,Readers,Online newspapers,Market Research,The Guardian 
I'm researching a piece for tomorrow's blog about the death of the internet (oh come on, can't you people take a joke!) and if you want to understand just how messed up and deceptive internet audience statistics are then I strongly recommend you try and do an exercise like this for your own newspaper.
I'll give you the conclusion at the beginning so you can decide whether to follow the tortuous mathematical route I had to take to get there.
Internet metrics substantially exaggerate the importance of the newspaper web audience. Would you be so keen to abandon print to its fate if you knew that despite 10 years of growth, your web site may only have 30% of the reach of your printed newspaper in the market where you can attract advertisers. And that at current rates of growth and decline it could be 2020 or 2030 before the two are even level with each other?
Newspapers are declining (for lots of reasons), the websites of newspapers are growing (slowly, if at all). Newspapers sure have some problems to tackle. But a business strategy that puts your web presence first is one that ignores which of those products reaches the most people. Print may be fast asleep. But it's not even close to being dead.
Some official statistics make the internet audience for newspapers seem very large indeed. Much larger than they actually are. And we print folk fall for it. We hear numbers with the word million in them and we didn't want to question it. But we have to now. Because this is the fuel that feeds the ideas of newspaper managements across the globe and makes them leap over the revenue precipice. And frankly, the numbers for print and web are tough to compare accurately like for like.
The most honest web people even admit the numbers are more propaganda than fact. As Simon Waldman said in January, the use of monthly aggregates was "a trend started in the early days of the web, when we needed to make our user figures seem big enough within our organisations".
He, with vastly more experience and access to data than me, found it a complicated exercise comparing newspaper website to newspaper website, let alone website to print.
But when you do strip it down, what you find is that despite big numbers being bandied around it is not safe to assume that your newspaper website is anywhere close to your newspaper in reach.
I picked the Guardian not because I believe they are the worst and most dishonest, but because I know they are the best and most transparent. (And because Simon has already done some of the work for me.) They also run a hugely successful site and a less successful newspaper. If their web metrics don't look good next to their print version then I dread to think how some others stack up.
Okay, here's the maths ...
1. I looked on the MediaGuardian website to find newspaper ABCs.
351k average daily sale for the Guardian over the 25 issues from 1 April to 30 April 2007.
The Guardian has 1.24m average daily readership, according to the NRS figures quoted in the Guardian AdInfo website.
2. I looked on Compete.com to find out the Guardian's unique visitors in a month.
Compete say 1.7 million in the US.
3. I looked on Alexa.com to find out what proportion of the Guardian's audience is in the US.
26% of total audience are in the US, 32% (2.1m) in the UK.
4. I looked on the Guardian AdInfo site to confirm the Compete and Alexa numbers.
The Guardian AdInfo site says 15.7m unique users in a month in the world, certified by ABCe.
They say 34% (5.3m) of these unique users are in the UK.
Oh dear. 2.1m or 5.3m UK unique users in a month? Which is right? I'm no statistician but even I know that can't be a rounding error. I must be doing something wrong.
Then I think I'm terribly clever and I've found the problem. When I looked at the ABCe certificate it appears that the way they calculate monthly unique users is to add up the daily ones during a month, a very newspaper way of looking at the world. So if I came to the site for 30 days in a row, I would count as 30 unique users. But I'm not clever after all and in fact I'm wrong. The daily unique user list adds up to 23,874,360. The certificate lists 15,170,438 unique visitors in the month.
But I'm still lost. 2.1m or 5.3m? Well, the Waldman piece suggests that Nielsen agree with Compete/Alexa on a UK monthly unique user number of 2.1m. Can't I just use that number?
Sadly I don't think I can. My point is to try and work out a comparison with print, so I'm going to take the 23,874,360 number. It repeats unique users from one day to another, just like an ABC number for a newspaper would.
Simon would agree. His January 2007 post suggests that daily unique users are "perhaps the closest we can get to a print ABC".
So that's settled then: 23,874,360 total daily unique users over a month is our starting point for web unique users.
So how to bring the data together to compare it?
The average daily number of unique users for the Guardian is (23,874,360 / 30) = 795,812
34% of them are in the UK, which is : 270,576.
Like-for-like means being brutal with the print figure and cutting out dubious "sales" of the paper that might inflate the figure unfairly in comparing it to the web reach. So I take away the estimated foreign sale of 42k and bulks of 15k and therefore reduce the headline Guardian ABC of 350,940 to 293,940.
So for my purposes the Guardian sold a daily average of 293,940 single copies in the UK in April 2007.
Using the same methods the Observer, the Guardian's Sunday newspaper whose content slots in to their website on the seventh day, sold 463,128 copies, minus about 68k foreign and bulks, for a total of 395,128.
So total sales for the month were (25 x 293,940) + (5 x 395,128) = 9.324 million.
Average daily sale was therefore 9.324m / 30 = 310,788.
So in the UK there were 310,788 people buying the Guardian or Observer on average each day in April 2007.
There were 270,576 reading guardian.co.uk online.
Pretty close eh?
Except I am not comparing like for like. We need to compare readers to readers, not readers to purchasers.
I hate the hocus-pocus of the UK's National Readership Survey but let's assume a conservative ratio of purchasers to readers of 1:3. (The Guardian AdInfo site quotes a ratio of 1:3.35.)
That would leave us the following.
UK Guardian and Observer daily average print readers: 932,364
UK Guardian and Observer daily average online readers: 270.576
A reality check for the maths.....
Total issue readership for the month = daily average print readership x 30 = 27,970,920
Total unique users in a month = 23,874,360 (x 34% in the UK) = 8,117,282
Yup, still 30%.
So the actual reach of the internet version of the Guardian in the UK where the two can be compared is about 30%. And that's a paper whose website vastly outperforms its print product!
And you want to use that to tell me that print is on its last legs?
Let's assume for a moment that this situation is typical of other newspaper sites. If you assume the web audience grows by 5% a year forever and the print product declines by 5% a year, it will still be 2020 before the two are even equal. If newspapers hold the audience steady and the web still grows by 5% it's 2030 before they get level.
Next time someone uses web metrics to tell you that print is dead I have some advice. Laugh.
I'll give you the conclusion at the beginning so you can decide whether to follow the tortuous mathematical route I had to take to get there.
Internet metrics substantially exaggerate the importance of the newspaper web audience. Would you be so keen to abandon print to its fate if you knew that despite 10 years of growth, your web site may only have 30% of the reach of your printed newspaper in the market where you can attract advertisers. And that at current rates of growth and decline it could be 2020 or 2030 before the two are even level with each other?
Newspapers are declining (for lots of reasons), the websites of newspapers are growing (slowly, if at all). Newspapers sure have some problems to tackle. But a business strategy that puts your web presence first is one that ignores which of those products reaches the most people. Print may be fast asleep. But it's not even close to being dead.
Some official statistics make the internet audience for newspapers seem very large indeed. Much larger than they actually are. And we print folk fall for it. We hear numbers with the word million in them and we didn't want to question it. But we have to now. Because this is the fuel that feeds the ideas of newspaper managements across the globe and makes them leap over the revenue precipice. And frankly, the numbers for print and web are tough to compare accurately like for like.
The most honest web people even admit the numbers are more propaganda than fact. As Simon Waldman said in January, the use of monthly aggregates was "a trend started in the early days of the web, when we needed to make our user figures seem big enough within our organisations".
He, with vastly more experience and access to data than me, found it a complicated exercise comparing newspaper website to newspaper website, let alone website to print.
But when you do strip it down, what you find is that despite big numbers being bandied around it is not safe to assume that your newspaper website is anywhere close to your newspaper in reach.
I picked the Guardian not because I believe they are the worst and most dishonest, but because I know they are the best and most transparent. (And because Simon has already done some of the work for me.) They also run a hugely successful site and a less successful newspaper. If their web metrics don't look good next to their print version then I dread to think how some others stack up.
Okay, here's the maths ...
1. I looked on the MediaGuardian website to find newspaper ABCs.
351k average daily sale for the Guardian over the 25 issues from 1 April to 30 April 2007.
The Guardian has 1.24m average daily readership, according to the NRS figures quoted in the Guardian AdInfo website.
2. I looked on Compete.com to find out the Guardian's unique visitors in a month.
Compete say 1.7 million in the US.
3. I looked on Alexa.com to find out what proportion of the Guardian's audience is in the US.
26% of total audience are in the US, 32% (2.1m) in the UK.
4. I looked on the Guardian AdInfo site to confirm the Compete and Alexa numbers.
The Guardian AdInfo site says 15.7m unique users in a month in the world, certified by ABCe.
They say 34% (5.3m) of these unique users are in the UK.
Oh dear. 2.1m or 5.3m UK unique users in a month? Which is right? I'm no statistician but even I know that can't be a rounding error. I must be doing something wrong.
Then I think I'm terribly clever and I've found the problem. When I looked at the ABCe certificate it appears that the way they calculate monthly unique users is to add up the daily ones during a month, a very newspaper way of looking at the world. So if I came to the site for 30 days in a row, I would count as 30 unique users. But I'm not clever after all and in fact I'm wrong. The daily unique user list adds up to 23,874,360. The certificate lists 15,170,438 unique visitors in the month.
But I'm still lost. 2.1m or 5.3m? Well, the Waldman piece suggests that Nielsen agree with Compete/Alexa on a UK monthly unique user number of 2.1m. Can't I just use that number?
Sadly I don't think I can. My point is to try and work out a comparison with print, so I'm going to take the 23,874,360 number. It repeats unique users from one day to another, just like an ABC number for a newspaper would.
Simon would agree. His January 2007 post suggests that daily unique users are "perhaps the closest we can get to a print ABC".
So that's settled then: 23,874,360 total daily unique users over a month is our starting point for web unique users.
So how to bring the data together to compare it?
The average daily number of unique users for the Guardian is (23,874,360 / 30) = 795,812
34% of them are in the UK, which is : 270,576.
Like-for-like means being brutal with the print figure and cutting out dubious "sales" of the paper that might inflate the figure unfairly in comparing it to the web reach. So I take away the estimated foreign sale of 42k and bulks of 15k and therefore reduce the headline Guardian ABC of 350,940 to 293,940.
So for my purposes the Guardian sold a daily average of 293,940 single copies in the UK in April 2007.
Using the same methods the Observer, the Guardian's Sunday newspaper whose content slots in to their website on the seventh day, sold 463,128 copies, minus about 68k foreign and bulks, for a total of 395,128.
So total sales for the month were (25 x 293,940) + (5 x 395,128) = 9.324 million.
Average daily sale was therefore 9.324m / 30 = 310,788.
So in the UK there were 310,788 people buying the Guardian or Observer on average each day in April 2007.
There were 270,576 reading guardian.co.uk online.
Pretty close eh?
Except I am not comparing like for like. We need to compare readers to readers, not readers to purchasers.
I hate the hocus-pocus of the UK's National Readership Survey but let's assume a conservative ratio of purchasers to readers of 1:3. (The Guardian AdInfo site quotes a ratio of 1:3.35.)
That would leave us the following.
UK Guardian and Observer daily average print readers: 932,364
UK Guardian and Observer daily average online readers: 270.576
A reality check for the maths.....
Total issue readership for the month = daily average print readership x 30 = 27,970,920
Total unique users in a month = 23,874,360 (x 34% in the UK) = 8,117,282
Yup, still 30%.
So the actual reach of the internet version of the Guardian in the UK where the two can be compared is about 30%. And that's a paper whose website vastly outperforms its print product!
And you want to use that to tell me that print is on its last legs?
Let's assume for a moment that this situation is typical of other newspaper sites. If you assume the web audience grows by 5% a year forever and the print product declines by 5% a year, it will still be 2020 before the two are even equal. If newspapers hold the audience steady and the web still grows by 5% it's 2030 before they get level.
Next time someone uses web metrics to tell you that print is dead I have some advice. Laugh.
This is a wonderful post. I have already cancelled Time and Newsweek to take the Economist. So now I should cancel my local newspaper and take the Manchester Guardian? (Probably has more news in it.)
Reply to this
Um... I wouldn't rely on Alexa as an accurate measure of traffic. First, it only measures the preferences of its toolbar users, and these may or may not be statistically representative of Guardian readers in the U.S.; as the Alexa toolbar is considered spyware by many academic, business and media IT departments, and is thus banned, I would guess that, at the very least, the amplitude is way off.
Reply to this
I tend to agree. Which is why I didn't rely on Alexa or Compete in the end. The 34% UK traffic comes from the Guardian. So does the 23,874,360.
I don't know enough about Alexa and how they weight their data, (though sampling and correcting large numbers of people is a well established technique and you'd hope they've thought about the issue) but I am intrigued that Compete and Nielsen came up with the same lower 2.1 million number.
But I can't explain it, so I didn't rely on it. But as data replicated by two sources it strikes me as potentially accurate. Maybe someone out there can tell me why it doesn't tally? Is it the problem of unique users not being the same as different people? Could that really double the results for ABCe?
I'll email ABCe and ask.
Reply to this
I must agree it is far too difficult to measure and compare effectively website visitors, sessions, impressions etc.
But I think it is important to remember that the Guardian can target its internet advertising geographically, something it can't do as effectively in the newspaper.
Until a simple, easy to use stats package is available that people trust it will be very difficult to get to the real numbers - oh if only ABC could see the potential if they led this from the front.
Reply to this
John, your math is wonderful and your work admirable -- right up to the point where you buy into the "one buyer equals three readers" mumbo-jumbo. I was with you right up to that point.
In some cases, that may be true -- in some cases, one buyer may equal zero readers. The one-to-three ratio is almost certainly bogus, but it's become gospel in the business because it makes newspaper reach look better to advertisers.
Reply to this
With all due respect, this blog post strikes me as a classic example of political speak. You set up a straw argument that is not being made and then spend a considerable amount of time debunking it. No one, that I am aware of, in the United States is concerned about readers fleeing newspapers for newspaper web sites. The problem is that readers and advertisers are fleeing newspapers - period. The business model for a newspaper was built on the printed page holding close to a monopoly on advertising dollars. If a business wanted to reach customers in a market, the business was forced to take out an ad in a newspaper. Radio and Television cut into that monopoly, but only by a small amount. The Internet has completely blown up that paradigm. There are now hundreds of thousands of publications competing for the advertising dollars. And while no single one of them can claim the size of the audience as a print paper, all of them can take a share of that limited advertising dollar leaving very little left for newspapers. In addition, in the past a newspaper had a way to attract a large and committed audience. It didn't matter what a person was interested in - be it sports, business, politics, government - there was one authoritative source of that information - the newspaper. Now, a person who is interested in sport can find a web site that focuses on that sport in ways a general interest newspaper cannot. Same with a person interested in politics, business, and the list goes on and on. This adds up to serious problems for the print newspaper industry. Readership continues to decline. Advertising dollars are continuing to flee. And there is nothing to indicate this trend will not continue.
Reply to this
Thanks for the comment. I totally agree with everything you say ... until the very end. I'd add a yet, and then I can agree with everything....
Your assessment of the mess newspapers are in seems pretty accurate to me.
And I share your view of the dismal chances for newspapers to thrive online.
In fact my straw argument is, I think, an important one. I believe newspaper people would be better focussed on seeking innovation in print than in migrating their product online and hoping for the best. Many many people disagree with me.
They say that the internet has destroyed the appeal of newspapers to everyone but old people and that therefore newspapers have to get online in a hurry. I'm seeking to challenge that, in my view, suicidally reckless viewpoint.
I believe we have to be aggressive in innovating in print if we are to survive. I guess I'm trying to convince a few people that the idea of total online migration as our only option is false.
Reply to this
What AmJourno isn't saying is that what's really hurting the US print market is that they're losing their classified-ads customers to Craigslist. Dunno if that's a problem for the Grauniad. (And really, most newspapers, even with the drop in classified ad revenue, are still nicely profitable; it's just that they want to retain the luxurious 20 to 30% profit margins of yore (profit margins bigger than what Big Oil demands) in order to please the shareholders, whereas figures of 5 to 10% are what's actually sustainable.)
Reply to this
True. And it is a big problem for the Guardian. But the problem is not insoluble if you're prepared to invest in real print innovation and revolutionise distribution.
My concern is that print companies believe that they can move online and get any profit margin at all. I'm no economist but with an effectively limitless free supply of content, no barriers to entry and millions of niche low-cost competitors alongside the numerous major ones, I don't think newspapers can even make 10-15% online.
I want newspapers to survive because I believe that they will offer an increasingly alluring alternative to what our future media landscape could look like. The explosion of information has been fantastic and well corralled by Google and the democratic possibilities of blogs are only just being unleashed. I love it all. But I also think that the gathered common ground, the ordered information, the permanence, the passive search, the tactility (tactiliciousness?) of print, the surprise that newspapers should offer can still be important if we adapt to the internet (stop being one product for everyone) rather than become its disciples.
Reply to this
A bigged up person at the Telegraph told me a few weeks after the Telegraph PM pdf was launched last year that offical numbers showed that it had a take up of about 600. "But I just don't believe those numbers because with the amount of visitors we get to the site everyday, you would expect to get at least 600 downloads from people clicking on the wrong part of the screen." That, in my view, is how internet statistics should be viewed. Nobody ever handed anybody else hard cash for a newspaper by accident.
Eminently yours,
Frank
Reply to this
Much of this is absolutely on target. I'm on record as having severely criticized my fellow onliners for tossing around meaningless and misleading monthly cumulative audience data, just as I've criticized the industry broadly for crowing about how print+online=growth when actual market penetration is measurably declining.
However, the notion that a newspaper's daily print sales figures should be MULTIPLIED by some factor to derive actual readers is wishful-thinking crap, and especially so in markets where the newspaper is home delivered, such as is typical in the United States. Try dividing! Once again, I ran over this morning's paper with my car on my way to work.
Readership declines are very real, and they're way ahead of circulation declines. Newspapers are getting tossed into driveways and front lawns, and left to collect dew and spiders until the next trash day.
Newspapers are severely abusing ABC rules on bulk sales, and many are carrying canceled subscribers on their books for as much as six months after being notified.
We don't have a medium-change problem so much as a content relevancy problem and a general failure to grasp the unique strengths of print and online and use each to best advantage.
We are not, overall, seeing a migration of readers from print to online consumption of news. If that were true, we should not be worried, as the economics of operating a news source online are actually fairly attractive IF you can get people to pay attention.
The much bigger problem is a general decline of interest in serious journalism, mingled with the rise of much better solutions for some of the jobs for which newspapers once were a preferred solution. (Selling a car, getting a job, and entertaining yourself would be obvious examples.)
Some of our problems can be fixed and some can't. The first step is to be honest with ourselves and to separate fact from our own PR spin.
Reply to this
I agree with Steve.
But moreso, I think any comparison is about as fruitful as comparing how many phone calls the Guardian gets compared to its readership.
If you are intent on comparing a print publication to your web services, you are on the wrong path completely.
They should be completely distinct and, if anything, complimentary, not competitive.
Reply to this
A good piece of investigative maths that proves web metrics have a long way to come.
First off daily newspapers all measure themselves on "average issue readership" which evaluates the number of people who looked at the paper for 2 minutes or more in the past day. The "time viewed" element is the critical part that is missing from any sensible use of web statistics. Until something comparable arrives in the standard web currency then it's all pretty subjective.
Secondly your maths uses the sandy foundations of current web statistics and uses the assumption that US users of The Guardian have the same density of consumption habits as UK ones is questionable. It would be fair to assume that the "home" readership of The Guardian would view more pages per session and have a longer dwell time that the US audience (if only due to cultural and sport content). This would suggest that your maths undervalues the UK component but overestimates the US one.
Reply to this