At the New Communications Forum conference last week, Chris O’Brien from the San Jose Mercury News mentioned during a presentation that 2/3rds of the traffic to the paper’s website comes from news aggregators (like Google News) and search engines.  This figure was higher than I expected, so I mentally filed away the tidbit to write about at some point.

Then today I came across a site called Newsknife, which breaks down which newspaper websites have the most articles appear in Google News.  I found this fascinating, so I quickly compared the twenty five newspapers that appeared most often in Google News with a list of the twenty five largest newspapers in terms of print circulation.  A couple of quick things jumped out at me:

• The Mercury News, Washington Times and Akron Beacon Journal were the papers that performed the best in Google News as compared to their print circulation.  For the Mercury News and Washington Times, I’d guess they do well because both papers have a lot of content niche news topics (silicon valley and politics) that are very popular on time.  So it makes sense that Google News might have a lot of their stories.  I have no idea why the Akron Beacon Journal is in Google News so much.
• The Wall Street Journal and Rocky Mountain News were the largest newspapers that did not appear on the list of top newspapers sites in Google News.  I suspect the Journal will start showing up soon given their recent deal with Google.  I don’t know what is going on with the Rocky Mountain News.

However, I was mostly left slightly baffled as to what to make of the data and why certain sites performed better than others.  How important is optimizing for Google News?  What is the impact of registration walls?  How important is the topics that are covered?  I think these questions are important to answers, as, based on the figures from the Mercury News, performance in search engines and news aggregators plays a huge role in the success or failure of a newspapers online program.  I’ll try to write more on this later after the data has a chance to sink in.

Anyway, below is the raw data.  Please share any thoughts you have in the comments.