Tumblelog by Soup.io
Newer posts are loading.
You are at the newest post.
Click here to check if anything new just came in.

May 04 2010

14:00

Moderating declines: Parsing the NAA’s spin on newspaper circ data

Newspapers could borrow a line from a recent Dilbert comic strip: “We’ve been doing great since we redefined success as a slowing of failure.” Or perhaps it was the other way around, and Dilbert creator Scott Adams was inspired to write that line in a recent strip by the inventive terminology of newspaper executives describing “sequential improvement” and “moderating declines” in their revenue trends despite continuing losses in the double digit range.

Currently, the industry is reporting first-quarter earnings, and last week the Audit Bureau of Circulations released unaudited “publisher’s statements” reporting paid circulation for the six months ending March 31. The numbers are down, but the spin is up.

On the circulation front, the Audit Bureau of Circulations reported that circulation fell 8.7 percent on weekdays and 6.5 percent on Sundays, among newspapers filing publisher’s statements. This compares with drops of 10.6 percent weekdays and 7.6 percent Sundays for the prior six-month period, enough of an improvement for Newspaper Association of America CEO John Sturm to declare that “the data indicates the declines are moderating.”

Actually, it’s hard to discern real moderation in the rate of decline. The losses in the most recent period are indeed a bit less severe than those in the prior (Sept. 30) period, but they are worse than the drop in the period before that, or in any previous period. If we ignore the Sept. 30 data as an outlier, we actually have a trend that’s been worsening steadily for the last six years:

Nothing about that final uptick indicates that it’s a reversal of the trend — it would take two or three periods of “improvement” in the form of “moderating declines” to make that a valid conclusion. In fact, both of the upticks in the trendline disappear if we take the statistically reasonable step of averaging spring and fall six-month circulation changes into annual figures and graphing those:

Moreover, viewed in long-term context, this latest minor slowdown in the rate of decline disappears entirely when newspaper circulation is viewed in the context of population: Since 1945, the number of papers sold per 100 households has dropped steadily, declining in 61 of the last 64 years.

(The circulation numbers on which this chart is based come from Editor & Publisher via NAA; E&P hasn’t released its Yearbook with a 2009 figure, so 2009 is my estimate based on the last two ABC cycles and estimated census households.)

It’s also quite possible that the uptick is entirely the result of some of the new options newspapers have in counting their circulation. Some of the declines of the past few years have come from ditching distribution in unprofitable outlying areas, and cutting back on “third-party” programs in which advertisers were persuaded to pay for bulk distribution, free to recipients, at community events or door-to-door in targeted areas. The value of this circulation was always questionable, but now ABC rules are permitting substitution of new forms of questionable circulation.

Take, for instance, the Bend (Ore.) Bulletin, where weekday circulation grew 34.3 percent. How? Since Jan. 1, with ABC approval, the paper has been counting e-subscriptions sold to current print subscribers for an extra 50 cents per month. As long as the subscriber can choose to opt in or out of the added digital subscription, the e-subscription counts as one paid subscription in addition to the printed one, even if the customer never accesses it. Applying this stratagem for only three months of six-month reporting period, the Bulletin tacked 12,462 weekday e-subs to its “core” print circulation of 29,072 (which is actually down by more than 1,000 from 30,155 a year ago). And next time around, counting the e-subs for six full months, it expects to report circulation of about 54,000.

How much of that is going on, and to what extent is it responsible for that uptick? I haven’t delved into the data, but Paid Content did, and reported that e-edition circulation was up significantly: the digital editions of the top 25 newspaper e-editions rose 40 percent, from to 1,363,212, versus 973,721 a year earlier.

There’s are other questionable figures being circulated, as well. In his statement on the ABC data, Sturm also cited readers-per-copy data that appears, at first glance, to mitigate the downward trend in copies sold: “Newspaper print products are also finding their way into more people’s hands, with readers-per-copy increasing by 7.5 percent in just the last three years to 3.3 adults on average, according to a recent analysis from Scarborough Research and Newspaper National Network LLP.” Here’s the graph:

Missing from this statement is the important qualifying statement that the Scarborough study applies to 25 selected “top markets,” not to all newspapers.

For its report (PDF download), Scarborough chose the 25 largest newspapers omitting “national” newspapers (New York Times, Wall Street Journal, and USA Today), as well as omitting papers in the midst of major circulation pattern transitions (Denver Post, Detroit Free Press, Philadelphia Inquirer, San Jose Mercury-News, Seattle Times and Seattle Post-Intelligencer).

In response to email inquiries, NAA’s research director Jim Conaghan and communications chief Jeff Sigmund defended Sturm’s statement. “He correctly cites the Scarborough research,” Conaghan wrote. “John Sturm’s statement references research conducted by Scarborough which was based on an analysis of the top 25 markets,” Sigmund wrote. But Sturm’s statement leaves out the important qualifier of the 25 markets. Conaghan also wrote to me: “You need to carefully read what is contained in the Scarborough report. Large markets/papers, survey data. The old NAA estimates were derived by a different method, and would represent total U.S.” By “old NAA estimates” he meant a 2007 report available at the NAA site that found 2.128 readers per copy weekdays and 2.477 on Sunday. (That study, in a footnote, discounts the validity of the Sunday number, stating: “Projection relatively unstable for Sunday RPC. Use with caution.”

Now, as that quote suggests, readers-per-copy (a.k.a. the “pass-along rate”) is a notoriously difficult thing to measure. But it has been tracked by Scarborough and others for a very long time and has been pretty consistently cited (and drilled into the heads of newspaper advertising representatives) as being around 2.3, plus or minus a point or two. This allowed salespeople for a 30,000-circulation newspaper to tell retailers that readership was actually about 75,000. The readers-per-copy factoid was included for years in NAA’s own “Facts About Newspapers,” a vest-pocket sized booklet of data distributed annually to advertisers and publishers. In 2000, Facts About Newspapers claimed 2.1 readers per copy, and 2.2 on Sundays. In 2004, it was 2.3 on weekdays, 2.4 on Sundays. In 2007, as cited above, it was 2.1 weekdays and (with a grain of salt in the NAA footnote) 2.5 on Sundays. When I started in the business, in the late 1970s, I recall that it was 2.6. In 1983, Scarborough and Simmons both came up with about 2.7, but that was noted as seeming to be on the high side. The average U.S. household is about 2.57 people, another reason why any reader-per-copy finding above that level is questionable. I think Conaghan is right on the money to say that Sturm’s claim of 3.3 can not be applied to any markets outside the those selected 25 and that reality is still in range of 2.2 to 2.5 as it has been for 50 years.

Sturm also pointed to the online newspaper audience, stating: “The latest Nielsen Online data found that newspaper websites attracted a record 74.4 million unique visitors per month on average in the first quarter of 2010 — more than one-third (37 percent) of all Internet users.” Left unsaid: the monthly UV average for this quarter was probably boosted in February by traffic related to the Olympics. The February UV count was 76.1 million, an all-time high. In March, with 10 percent more days than February, UV’s were 5 percent lower at 72.1 million.

But everyone knows by now that UVs are not a very good indicator, and “time spent on site” is what counts, at least if you’re trying to sell advertising. By that score, Q1 did not shape up very well for newspapers: in January, the average visitor spent 33:09 minutes at newspaper sites, in February, 29:06 minutes, and in March 32.21 minutes. These are three of the four shortest attention spans recorded by Nielsen Online for NAA since January 2004 (with the caveat that data before June 2009, which showed significantly higher times spent, was based on a different survey sample and can’t be compared with later stats).

Here’s the trend line for attention, or time spent, at newspaper websites since the June 2009 methodology change.

When that trend starts to demonstrate clear and strong “sequential improvement,” newspapers will have something to shout from the rooftops.

April 05 2010

16:43

Is print still king? Has online made a move? Updating a controversial post

A year ago, in a Nieman Journalism Lab post that garnered 88 comments and still has viral life out there, I maintained that just three percent of newspaper content consumption happens online; the rest of it happens the old fashioned way, by people reading ink on dead trees. Given the continuing attention being paid to that conclusion (it was cited just last month by Hal Varian, Google’s chief economist, in testimony to the Federal Trade Commission), let’s revisit the numbers and see whether anything has changed.

With updates or improved data on at least some of the numbers, the general conclusions still hold: U.S. newspapers have not pushed much of their audience to their websites, nor have they followed the migration of their readership to the web. Their combined print and online readership metrics, whether measured in pageviews or in time spent, show that there’s been significant attrition since last year in the total audience for newspaper content, and that the fraction of that audience consuming newspaper content online remains in the low-to-mid single digits.

Here’s how I arrived at the numbers this year (to follow this more closely, or check my math, you can view my worksheet here):

Point of comparison: Pageviews

First, a comparison of pageviews in print and pageviews online. In print, I projected pageviews for newspaper content by taking the 2008 paid circulation reported by the Newspaper Association of America, adjusting it by the average of the two six-month circulation loss figures reported by the Audit Bureau of Circulations (March, September), and multiplying the resulting 2009 circulation by 2.128 readers per copy for weekdays and 2.477 for Sundays. (This is a 2007 Scarborough Research (PDF link) number I used last year also, but readers per copy has been a very consistent figure with little variation for decades.) This yielded total readership for weekdays and Sundays. I then made the same (discussable) assumption as last year: that the average reader of a newspaper issue looks at 24 pages, which means there is a total of 70.602 billion printed newspaper pageviews per month. That’s down almost 19 percent from last year’s 87.1 billion pages viewed. To be fair, my audience numbers last year were also based on that 2007 Scarborough data, so that’s really a two-year decline. (Jim Conaghan, research director at the NAA, tells me they have no data on the number of printed pages readers look at on average, and that there is no update to the 2007 readers-per-copy study.)

For online pageviews, NAA offers a precise number based on research by Nielsen Online. Nielsen’s methodology changed in June 2009, so I’ve used the average of the nine months from June 2009 to February 2010, which was 3.382 billion online newspaper site pageviews per month. So for print and online combined, we have a total of 73.985 billion pageviews (versus 90.3 billion last year). In other words, as measured in pageviews, 95.43 percent of total readership for newspaper content was in print; 4.57 percent of it was online. So while it appears that the online fraction has grown from 3.5 percent in the previous analysis, the bad news is that the total content exposure has dropped by about one fifth.

Point of comparison: Time on site

Some commenters to last year’s post maintained that print and online pageviews weren’t comparable. And certainly, the current wisdom says that pageviews and unique visitors don’t count nearly as much as “engagement” as measured by time spent on site as well as interaction with content. So, as I did last year, let’s look at time spent — both in print and online, print engagement versus online engagement with the newspaper content:

For the print side of the ledger, I began with the readership counts derived as above, and assumed average time spent with printed newspapers to be 25 minutes on weekdays and 35 minutes on Sundays. Now, this assumption got considerable comment flak last year, and no doubt will have its doubters this year. For those who say “I don’t know anybody who reads a newspaper at all, so how can the average be 25 minutes?” let me say that more than 40 million newspapers are still sold every day and someone is reading them, whether you know them or not. Anecdotally, half the people I see at Amy’s in Brattleboro are spending more time that that just with the New York Times. But let’s avoid the anecdotal evidence — here’s (PDF link) some U.S. Statistical Abstract data on time spent with various media, sourced from Veronis Suhler. It claims that the average person in 2009 spent 159 hours a year with newspapers (including newspaper websites), which is 26.1 minutes a day. While this tends to support the controversial pass-along factor, it’s for the average (adult) person. Since only about half the population actually reads printed newspapers (on average per day), that would mean newspaper readers spend an average of 52 minutes a day — which just strikes me as way too high. So I’m going to stick with the happy medium of 25 minutes weekdays and 35 on Sundays until someone can improve that data. (As an additional data point: According an NAA print newspaper “engagement” study (PDF link) presented a few years ago, on weekdays 45 percent of readers spent more than 30 minutes, 34 percent between 16 and 30 minutes, 21 percent under 15 minutes. Higher times were reported for Sunday editions.)

That yields total time spent with printed newspapers of 78.471 billion minutes per month. The online side is easy: averaging the last 9 months of NAA data, we get time spent at newspaper websites of 2.535 billion minutes per month. And combining print and online time spent, we have a total of 81.006 billion minutes per month spent with newspaper content. The engagement measure, therefore, says that 96.87 percent of time spent with newspaper content was in print; 3.13 percent of time spent was online. This is almost exactly the same as last year, when I found that 3.0 percent of time spent was online. But printed newspapers have lost a big chunk of total engagement as well: this year’s numbers are down 18.9 percent from last year’s analysis, which, again, really is a two-year drop of about one-fifth, with the loss occurring on the print side.

The conclusion that the overwhelming share of newspapers’ audience remains on the print side of the ledger is supported by Scarborough’s 2008 ratings of what it called the “Integrated Newspaper Audience” (PDF link) in selected markets. Measuring the cumulative 5-day audience rather than daily averages, that data showed that the incremental audience at newspaper websites added only a few percentage points to their print reach.

NAA and Nielsen are clear that their pageviews and time-spent stats since June 2009 can’t be compared with earlier months because of methodology changes, so I’ll refrain from doing that; but clearly the print/online audience split was enormously skewed last year and remains so — and most importantly, the online side is not growing. Back in June, NAA reported 3.469 billion pageviews and 2.701 billion minutes spent; in January (to avoid the short month of February), there were 3.452 billion pageviews and 2.485 billion minutes spent. Time spent per unique visitor has fallen gradually from 38:24 minutes in June to 33:09 minutes in January. In other words, while newspapers are losing readership on the print side, that disappearing audience is not following them online; at best, the online audience for newspaper content is static.

The purpose of this analysis is not to compare all “offline” news consumption with all online news consumption; it is to dissect the newspaper content audience. But as several commenters noted last year, this really means that as the audience moves online, it is getting most of its news from non-newspaper sites.

Beyond examining the split between readers of printed and online newspaper content, I also noted in another post last year that newspaper websites attracted less than one percent of all U.S. web traffic — 0.69 percent of pageviews and 0.56 percent of time spent, to be precise, in June 2009. Updating those stats with February 2010 Nielsen Online data (also detailed in the spreadsheet linked above), over the last nine months newspapers have actually lost share in both pageviews and time spent: pageview share dropped to 0.63 percent, and time spent dropped to 0.50 percent of total web traffic.

Meanwhile at newspapers, much effort and much dialogue continues to focus on getting readers to pay for content and battling aggregators — energy that might better be spent figuring out how not to lose the sizeable remaining audience for newspaper content, not by “protecting print” but by keeping the current print readers in the fold as they, too, gradually migrate to reading news online.

14:15

Three iPad design choices that will influence how we read news online

So we don’t have to guess about what news apps on the iPad will look like any more. With Saturday’s debut of the device — which is, oh by the way, amazing — we now know how about a dozen major news organizations have chosen to present themselves on Apple’s new platform. I think it’s fair to say that we’ve seen no revolutionary apps to this point — solid, competent, but not revolutionary — but that doesn’t mean there aren’t already some important lessons to be learned from what’s already out there. After a weekend of playing around with all the news apps I could find, here are three design choices that I think are worth taking a closer look at.

Story-to-story navigation

Web-analytics types push time-on-site and pageviews-per-visit as measurements of a reader’s engagement. It’s fine to have someone following an occasional link in Twitter, but the real money, some argue, is in dedicated readers who spend lots of time with your content. And to that end, a number of sites have been working on their internal website navigation to push users from story to story, rather than asking them to head back to a list of headlines first.

On news iPad apps, that story-to-story navigation has become the norm. The Wall Street Journal, BBC, Associated Press, USA Today, NPR, Reuters — their apps all allow (and in some cases emphasize) swiping or tapping from story to story rather than Back and Forward, web browser-style. (The New York Times’ app is an interesting exception.) I suspect that’ll lead to more stories consumed per session — and I wouldn’t be surprised if you didn’t see more news companies taking that lesson back to their websites.

Diving right in

Similarly, just about every news website created in the past 15 years has pushed users down a similar path: show them a whole bunch of headlines, arrayed into a variety of design styles, then expect the user to choose one of them and begin what the site hopes will be a lengthy run of clicking on stories. It’s a decision tree: Here are your options, now make a choice.

The very attractive BBC app takes a key step away from that pattern. When you launch the app, you’re not confronted just with a bunch of headlines — you’re also thrown immediately into the text of the app’s top story, without so much as a click. And once you’re reading one story, the act of flicking to another one seems closer to a default act than when you’ve just selected from a menu of options.

It’s a model that makes perfect sense from a broadcasting background; a BBC radio or TV show doesn’t wait to ask which story the listener wants first. It just dives right in. Considering how many news website users never get past that list of initial headlines, dumping the reader directly into a story might be a way to push browsers into readers. The BBC may not rely on in-app advertising to pay the bills, but for sites that do, it’s a model worth watching.

The Times’ cyberclaustrophobia

There’s been a hearty debate among a certain breed of techie over whether the iPad’s existence as a closed ecosystem of apps makes it somehow evil. I think that argument’s more than a little overblown, but using The New York Times’ Editors’ Choice app, I was surprised at my own reaction to the closed universe of Times news it presents.

The free Editors’ Choice app includes a limited sampling of Times stories; presumably, that’s because the full New York Times experience is being held back for a subscription-only app to come later. But the experience of using the app is markedly different from going to nytimes.com, because it has an endpoint. Give yourself two minutes and you can easily read every headline, with five swipes and four taps. Not too much longer and you can read all the stories you’re interested in reading.

On one hand, I’ve heard print-centric people complain that one reason they prefer newspapers over their websites is that, with a newspaper, you know when you’re finished. Turn that last page and you know your news-collection job is complete. With a news website of any size, there’s always one more place to click, always one more link to follow, always one bit of breaking news that dribbled in since you started reading. Maybe an approach like the Editors’ Choice app’s can give a jolt of emotional satisfaction to those users.

But for me, the Editors’ Choice app just gave me a weird case of cyberclaustrophobia. It’s like swimming in the river of news only to find it ends not at a lake but in a parking lot. It’s like reaching the end of the Internet.

I can’t knock the Times, or pretend my reaction is logical. They are giving away around 50 stories at any given time, nytimes.com is still a click away in the iPad’s web browser, and I’m glad to see them (and others) experimenting with paid-content strategies. But at a gut level, reaching the end of a Times digital project still feels wrong after more than a decade of training.

March 04 2010

15:00

The Newsonomics of time-on-site

[Each week, our friend Ken Doctor — author of Newsonomics and longtime watcher of the business side of digital news — writes about the economics of the news business for the Lab.]

Parse out the numbers, and they’re quite puzzling.

The average news reader spends little time on newspaper-owned sites, from a 20 minutes a month or so on the New York Times site to eight to 12 minutes on most local newspaper sites. That’s minutes per month. Those numbers, as tracked by Nielsen and reported monthly by Editor and Publisher, are steady at best, showing, in fact, some recent decline. They are, literally, stuck in time.

Then, take the number of minutes Internet users spend on social sites. Nielsen’s January tally showed seven hours of usage a month on Facebook alone, in the U.S., blowing away all competition. That’s some 40 times more time spent on social sites than on any single news site.

Which is a bit deflating for those in the news business. So let’s try to get at what the numbers may be telling us.

Maybe that big Facebook number isn’t as important as we think. We all have long spent much more time in conversation, much of it idle, some of it about what we’re doing right now or plan to do (the “statusphere” of the pre-digital world) than we have in reading the news. So social-site time may replace water-cooler conversation time. Further, do those Nielsen numbers mean that someone is actively perusing Facebook walls (or Twitter feeds) until their eyes fall out — or that they are keeping windows open on their computers? Are they engaged in a way that advertisers care about?

Then again, if Facebook time is a proxy for our new information centers — where we go to find out what’s happening in the community and the wider world — then it is becoming the new home page. Recall how newspaper sites all put up “make us your home page” buttons more than a decade ago? Constructively, that’s what Facebook done, without the button. That’s not surprising; it’s the ultimate page about what we care about most: me. Sure, some of the posts tell us about the wider world, but a good 80 percent or more tell us something personal.

If social sites, including Twitter, are a new center — Nick Negroponte’s “Daily Me” morphed — that’s a new challenge, and maybe opportunity, for the news industry. The challenge: getting the news to where the readers are hanging out, and figuring out to monetize there. The opportunity: If properly seeded in the social sites, the readers themselves do the (free) marketing and distribution of the content. The early tests of Facebook Connect appear promising here, though too few news companies are experimenting at any kind of scale. (See “The Newsonomics of social media optimization“.)

Now, let’s look at the Newsonomics of time-on-site — how well such time is monetized.

We’ll do some extrapolating with Facebook, to figure out what 2010 might look like. Let’s start with January numbers of 113 million U.S. users and seven hours time spent. Let’s be conservative and say for the year, it ends up with 120 million users and the same seven hours. That’s 84 hours a year for the 120 million, or a little over 10 billion hours of time spent.

For newspapers, let’s use one of the higher-achieving companies for comparison. The New York Times has been averaging about 20 million monthly uniques. It’s time-on-site varies considerably, with the news (!). Let’s give it 25 minutes a month on average. That’s 5 hours a year, or in total, about 100 million hours.

So, in time spent, the Times is less than one percent of Facebook.

Now, let’s look broadly, and quickly, at revenue. The Times’ 2009 digital revenue: about $342 million. Or $3.42 for each hour spent on the site.

Facebook’s revenue numbers are unannounced, but smart industry speculators put its 2010 number at about an even billion dollars. Or about a dime an hour of time spent.

$3.42 vs 10 cents. The Times is monetizing its time on site 34 times better than Facebook.

The Times and other big established news brands will say that’s more than fair, given the attention of the audience, the premium nature of the content and the demographics of the audience. Facebook, and its financial and spiritual advisors, will tell you that’s all upside. They’d point to yesterday’s partnership announcement with (Adobe’s) Omniture on ad placements as just one small step to a large revenue future.

Photo by Robbert van der Steeg used under a Creative Commons license.

Older posts are this way If this message doesn't go away, click anywhere on the page to continue loading posts.
Could not load more posts
Maybe Soup is currently being updated? I'll try again automatically in a few seconds...
Just a second, loading more posts...
You've reached the end.

Don't be the product, buy the product!

Schweinderl