Much buzz last week about the release, via E&P, of Nielsen’s May time-spent ratings of the top US 30 newspaper sites, and the fact that 17 of them — more than half — showed a decline from year-earlier levels.

Now, any statistician will tell you not to read too much into one month’s data. It’s like drawing conclusions from one day’s fluctuation in the stock market. I’d bet the ratings were up, big time, in November and January on Presidential election and inauguration news. In May 2008, among other things, there was a lot of attention being paid to the final primaries and pre-convention maneuvering in the Presidential race — far more interesting news than we had last month. Clearly, that explains why time spent at Politico dropped about 4 minutes from last year’s 10 minutes, 21 seconds. Barack Obama clinched the nomination on June 3, 2008; Hillary Clinton threw in the towel on June 7 — so when the June 2009 ratings come out, we could see another lackluster month, unless traffic is boosted by events in Iran.

Also puzzling: in some cases, while time spent is plummeting, unique-visitor counts are soaring.  For example, the May data has time spent at NJ.com, the Star-Ledger’s site, dropping from 8 minutes and 25 seconds to 2 minutes and 28 seconds.  Meanwhile, however, according to compete.com, uniques at NJ.com have soared from 1.2 million in May 2008 to 2.7 million last month, with site visits zooming during that period from 4.2 million to 7.5 million.  Both UVs and visits show a pretty consistent uptrend especially since November.

What’s going on there?  Back in August, the site underwent a redesign promising “to make its wealth of in-depth content easier to find and use.”  Perhaps, too, there has been more local promotion of the site.  But all those extra eyeballs are obviously not sticking around very long.  And when investor-oriented 24/7 Wall Street rated the top US newspaper sites last week, it singled out NJ.com as deserving an F rating, with this description:

The paper’s website, NJ.com, is the worst one in the survey. It is confusing from the start ….  Whatever discretion the editors of the Star Ledger had to deviate from the site’s cookie cutter formula, they used to the detriment of the reader. The “Business” section lead-in on the home page is simply a poll about wedding guests. No one would guess by looking that the stock market or economy even exist. The only really intelligent decision that the editors make on the front page is to put video content near the top and allow people to submit their own video content, a smart way to promote interaction. Matters get worse at the paper’s inside sections. Photos look like they were taken with disposable cameras. The sports section has the appearance of one long blog. Almost all of the stories run based on what time they were filed. Some of the most important stories of the last day disappear to the bottom of the page. That makes story selection mechanical instead of editorial. The same madness carries over to other sections like “Business” where the reader could use some help determining the priority of the news. NJ.com is probably not prospering. The most prominent ad is for colon cleansing, which is ironic given the quality of the paper and the way is set up so it will not betray any human hand or intelligence.

When 24/7’s ratings came out, I tried to match their grades with traffic growth rates and other metrics via Compete and others, but could find no clear correlation with the grades.  NJ.com got an F but shows a nice traffic uptrend.  Now I’ve gone back and matched the grades to Nielsen’s time-spent data, and guess what?  It still doesn’t make any sense.  Here’s the list of those papers/sites appearing on both lists* ranked from highest 24/7 score to lowest:

24/7 letter grade, Paper/site, 2009 time spent, 2008 time spent

A New York Times 27:34 28.52
A-  Newsday 3:30 3:35
A-  St. Petersburg Times 5:41 7:59
B+ NY Daily News 5:47 6:16
B Atlanta Journal Constitution 16:13 11:11
B Chicago Sun-Times 5:47 8:09
B Chicago Tribune 9:13 8:20
B Detroit Free Press 17:33 12:48
B Minneapolis Star Tribune 47:25 27:18
B-  LA Times 7:37 7:13
B-  San Francisco Chronicle 21:00 12:01
B-  Washington Post 10:58 16:04
C NY Post 10:57 8:02
D+ Houston Chronicle 24:42 21:43
D Boston Globe 10:06 9:40
D-  Dallas Morning News 4:43 4:51
D-  Philadelphia Inquirer 4:46 8:03
F Star Ledger F 2:28 8:25

If you can find any consistence between grades and times, let me know.  The New York Times has strong and consistent time spent matching its A rating, but Newsday and St. Petersburg’s tampabay.com, with A- grades, are in the single-digits in time spent. At the bottom of the list, the Star-Ledger’s F matches its dismal time spent, but the Houston Chronicle gets more than 20 minutes despite its D+.  For the group that earned solid B grades, the time spent ranges from the Star Tribune’s impressive three-quarters of an hour all the way down to less than 6 minutes at the Sun-Times.  And of the 12 sites with B- or better, 7 lost time year over year, about the same ratio as the whole top-30 group listed by E&P.

So clearly, subjective ratings such at those by 24/7, professionally and carefully done as they were, get an F in predicting engagement measures.  I’ll return to this topic in a day or two with some thoughts on where we might look instead for factors that predict how much time people spend with newspaper sites.

*Nielsen includes the top 30 ranked by unique visitors; 24/7 Wall Street graded the top 25 ranked by circulation; the lists overlap by the 18 listed newspapers/sites.

Photo by badboy69, used under Creative Commons License.