Martin Langeveld has been doing some interesting stuff over at Nieman Journalism Lab on newspaper numbers. I disagreed with parts of an analysis of his the other day, but it was still well worth trying.

Today, he does another rough calculation to find that just 3.5 percent of newspaper reading is done online. Nearly 97 percent of reading is still done with the fish wrap. At first glance, it doesn’t look right, but Langeveld’s using industry numbers along with a couple of fair assumptions.

I think this is important because it reminds us of how critical newspapers remain for tens of millions of people’s everyday lives. We hear the terrible industry news constantly and kind of assume that nobody reads them anymore. But more than 40 million people pay for newspapers every day via subscriptions or newstands. The bonus for advertisers: More than two people read each of those copies.

The NAA’s research shows a “daily” (Monday through Saturday) print audience of 116.8 million, and a Sunday print audience of 134.1 million…

We don’t have clear data about the average number pages each member of that audience looks at, but let’s make an educated guess: 24. That translates to about 87.1 billion printed page views per month*

Now for online:

For 2008, it averaged 3.2 billion online page views per month. (There’s no readers per copy multiplier there, on the assumption that nearly always, there’s just one pair of eyeballs per online page view.)

So, U. S. daily newspapers deliver a total of 90.3 billion page impressions per month, print and online. The online share of these page is only 3.5 percent — 96.5 percent of page impressions delivered by newspapers are in print.

I think where this may be off is that I suspect the online page views are much higher than those calculated by Nielsen, which uses a survey for its numbers. I just talked to the Journal the other day about why Nielsen’s numbers gave it 6.8 million unique visitors a month when its editor said WSJ Online had 23 million uniques a month. The Journal said its numbers come from its site monitor, which measures actual traffic, not by surveys and extrapolation. I know our traffic at CJR, for instance, is many times higher than estimates online.

Still, using the Journal example, let’s say the Nielsen online numbers are only about one-third of the real number. That still leaves print with a nearly 9-to-1 lead over the Web, which coincidentally enough, is about the ratio of print revenues to online revenues.

Langeveld’s post has much more in there, including calculations of advertising CPM’s (cost per thousand views), that raise interesting points and red flags. Using Newspaper Association of America numbers, he calculates that Web ads’ CPM is $80.28, which is ridiculously high. The real cost is less than 10 percent of that number.

As he says:

(That should raise your eyebrows, because if there are maybe three ads on the average page, it means the average ad is selling for more than $25 per 1000 views, which would be off the charts for most sites. I don’t buy that number, but that question is the subject for more research and hopefully a future post.)

On the print side, NAA reports 2008 revenue of $34.74 billion. Dividing that by 12 months and 83.6 billion printed pages per month, we get a print CPM of $34.62.

So how did that online number go awry? First of all, it bolsters the case that the 3.2 billion pageviews per month is way too low. But it also could imply that the $3.1 billion in 2008 online revenue reported by the NAA is way too high.

I suspected this the other day when I did a back-of-the-napkin calculation on how much revenue The New York Times’s website brings in a year. I came up with somewhere between $130 million and $170 million (I got to this by backing some numbers out of the Times Company’s 2008 securities filings, so they’re a very rough estimate).

Ryan Chittum , a former Wall Street Journal reporter, is deputy editor of The Audit, CJR's business section. If you see notable business journalism, give him a heads-up at rc2538@columbia.edu.