facebook sdk

google tag manager

Tuesday, April 11, 2006

Feed Analytics: Case Study #1

I'd like to explore the current state of feed analytics in MSM, looking at it from the outside in. I do not have access to the analytics for any of these media sites - I'm only taking a best guess here based on the data collected as a consumer of these feeds. If anyone reading this has any further information, or clarifications, I'd love to hear from you.

First up is Newsweek, which is a MSNBC property, hosted on MSN. One of the reasons I've chosen Newsweek first is that they are one of the MSM sites with feeds hosted by FeedBurner, and is one of the first media sites to leverage FeedBurner's FeedFlare tools. FeedFlare gives you the ability to add footers to each feed, providing a little extra functionality to each article. It's a brilliant tool. So, what is Newsweek using FeedFlare for? Take a look at one of their feeds (they have 40 of them!)...and you'll see they are including:
  • Email this (essentially a "mailto" link)
  • Technorati (if there are technorati links to the article, this flare appears - and links to a technorati search for the article)
  • Add to del.icio.us (posts this article to your del.icio.us bookmarks)
How does FeedBurner do all of this dynamically for each article within each feed? Each of the FeedFlare items is an image of the text with a dynamic URL that gives FeedBurner the ability to track which FeedFlare link was clicked (to give clickthrough statistics). They've expanded this service recently - and when they allow for dynamic flares, then things will really get interesting!

Another thing you'll note when you visit a feed is the Newsweek logo in the upper right. This is a standard feed image for RSS. Every time the feed is read (given that the RSS reader reads images), the image is called from:
  • http://msnbcmedia.msn.com/i/msnbc/Sections/Newsweek/NWProjects/NW_RSS/rss_logo.gif
Newsweek could obviously analyze the traffic to that image request. That would give them information about how often the feed is viewed (feed impression). Of course, FeedBurner will give them that stat as well.

Many of the articles have images. Each image in the feed is pulled from MSN too, so you could analyze those images to see when articles are viewed (article impression). Again...FeedBurner will provide that stat too.

If you do decide to click on an article, here's what happens:
  1. The actual link is to feeds.newsweek.com, which is a CNAME to feedburner.com
  2. Which redirects to a specific article ID in FeedBurner (example: http://feeds.feedburner.com/newsweek/TechnologyScience?m=55)
  3. This redirects to the Newsweek site for the article with "/from/RSS/" tacked on the end of the URL which can be picked up in their analytics package
It's a great implementation of feeds and of FeedBurner. They have branded the feeds (the URL is to feeds.newsweek.com), and embellished the feeds (via FeedFlare) to best take advantage of this maturing technology.

To summarize the analytics, I believe they have the following impressive set of analytics data available:
  • Feed impressions (via FeedBurner)
  • Article impressions (via FeedBurner)
  • Article clickthroughs (via FeedBurner and via the URL modification)
  • Various other feed stats that FeedBurner provide, including:
    • Types of feed readers used
    • Number of subscribers to the feed
    • How often a feed article is emailed, searched for on Technorati and bookmarked into Del.icio.us
Additional information that requires some work to tie together:
  • Returning visitors to the feed - they would need to analyze the traffic to the Newsweek feed image, and tie the visitors back into their analytics package
  • Returning visitors to individual articles - where there are images in the article, they can analyze the traffic to those images. Where they don't have an image in the article, there is no reference back to Newsweek for that visitor.

1 comment:

  1. Hi, I am extremely far behind on my blog reading so sorry for the late comment. I also work for a web analytics vendor and have performed some research into feed analytics as a kindof pet project. As of right now, I consistantly recommend Feedburner for feed analytics because tracking image requests from feeds does not give an accurate representation of article impressions. Just like xml/rdf requests or header image loads do not give an accurate representation of subscribers. There is just too much variation in the way feeds are read and the way readers behave. The article I've read that gives the most holistic view of feed analytics today to my mind is Measuring Your Blog and RSS Success.



webtrends reinvigorate analytics