The Cash Is In The Topics

All conversations I keep having about the economics of a news web sites revolve around two key ideas: how to increase both the duration and the depth of a visit. In this respect, much work remains. For August 2009, here are the numbers of page views, as measured by Nielsen Net Ratings on the French market. :

  • Online gaming:         between 400 and 600 page views per person and per month
  • Facebook:………………………411
  • Google:………………………….260
  • Meetic (dating site) :………..208
  • Leboncoin (free classified):..182

That was for the top 17 French sites ; further down in the rankings, medias sites go like this:

  • TF1 (n°1 television network):…39
  • Le Monde (national daily):………24
  • Ouest-France (regional daily):…22
  • 20 minutes (national free):…….20
  • Le Figaro (national daily):………20
  • Les Echos (business daily):…….14
  • Rue 89 (pure player):……………10
  • LePost (pure player):………………8
  • Liberation (national daily):……….8
    (Those numbers apply widely elsewhere as behaviors don’t vary much from one market to the other)

Well. You see where I’m going: if you set aside tricks such as massive video or slide-show contents used to artificially increase page view numbers, heavily used news sites such as Le Monde’s or Le Figaro’s get less than 5% of the monthly page views of a gaming site. Agreed, you shouldn’t compare gambling and consuming information; but in analog life, there isn’t the wide difference we observe in the digital universe, on line, between the amount of time people spend playing the lottery or betting on horses, and reading a newspaper or a magazine. In other words, the internet hugely accentuates the viewer’s “engagement” in favor of social entertainment, at the expense of consuming solid, but static contents such as news. Depressing indeed. But here is the good news: there is room for improvement.

If we take 4 pages per visit as a credible average for a news site, adding 2 more pages  per visit will sharply increase the actual advertising revenue per visit. Not by 50%, of course, since the more you add pages, the less valuable they become (CPM drop fast once you leave the hottest part of a site), but still worth the effort.

A few months ago, I mentioned the importance of recommendation engines comparable to the ones used by e-commerce sites. These engines can be algorithm-based — people who like this, like that — or socially-based such as ratings systems. Speaking of which, I don’t mean to ruin the delicious mystique of the crowd’s wisdom, nor do I intend to demean the multitude’s ability to unearth the good stuff and dismiss the bad; but, as this story in the Wall Street Journal shows, internet users tend to be generally too kind: a survey of 600 sites shows the average rating is 4.3 stars out of 5. This exuberance makes most of the rating process pointless. Better find something else.

Like the semantic web.

Last week, at the Berkeley Media Technology Summit in California, Thomas Tague, from Thomson Reuters, presented the latest developments of the OpenCalais project. In a nutshell, it goes like this:

  • You have a corpus of text(s), such as the stories in your publication
  • You run the whole body of content through the OpenCalais system, which extracts the following:
    • People’s names
    • Companies, organizations, publications
    • Places (all sorts)
    • Events (e.g. company X is launching product Y)
    • Quotations (great for journalists and scholars)
    • Thesaurus (to a certain extent, extracted words such as industry terms, can be re-arranged in a catalog of meaning groups)
  • In addition, it connects the generated tags to sources in a position to provide further depth. These could be mainstream sources such as the various flavors of Wikipedia, Crunchbase (the business database of TechCrunch), Reuters, International Movie Database or any kind of preset databases (online retailers, catalogs, product reviews, news archives, governmental databases, etc).

To get an idea, paste text of your choice in the OpenCalais viewer here. It will generate dozens of tags that will, in turn, be linked to other sources. Pretty spectacular.

All of a sudden, a corpus of relatively inert text, becomes a set of elements that can be searched, retrieved and connected to the rest of the internet (or to your own stuff). Needless to  say, as long as any content — photo, video, infographics — is properly tagged, it can be networked to the tags generated by Calais. For instance, the Calais plug-in I installed on the Monday Note site (Tagaroo), is linked to Flickr; in practice, each tag generated by Calais gives access to a vast library of pictures available under the Creative Commons license. Quite convenient (think about the looming standardization of contents — another subject).

A question you might ask: How does this technology improve my website’s profitability? Well, one of the key differences between a real news site and a shallow blog is the depth of its data inventory and, as a corollary, the potential for a great deal of added value around themes and topics. As a example, The New York Times has been able to greatly increase its viewers’ stickiness by adding a Topic page to almost every significant piece of news (they home-brewed their system). And the NYT didn’t spare its efforts: no less than 14,000 topics have been created with archives going as far as 1981!

At the Berkeley conference Thomas Tague mentioned several cases of big websites who benefited from Calais. CBS Interactive’s Cnet, The Huffington Post or the New Republic, have seen the following improvements:

  • Better search engine placement (search results land higher in Google)
  • Ability to localize their content through microsites targeted to a particular community
  • Cost reduction through more effective content production (once every single piece of news is structured, it is way easier to organize contents)
  • Ability to create specialized “niches” based on certain topic, sector, special interests, perspectives, etc.

All of the above having a positive impact on the critical metric: the number of pages per visit.

Launched in January 2008 by Thomson Reuters, OpenCalais has grown quite a lot. Built on the principle of a free and open system, OpenCalais now enjoys the support of a community of 15,000 developers. When asked why this tool is made freely available by Thomson Reuters, Thomas Tague replies that “surviving as ‘walled garden‘ of contents” is an outdated concept.

One last thing. I asked Thomas Tague about the name “Calais” (the closest French city across the Channel); he replied that, when, in 1851, Julius Reuters decided to switch his business model from flying pigeons to telegraph, he relied on the first submarine cable than landed in Calais. That was the beginning of the Reuters we know…

Related columns:

  1. Cash isn’t Cash Anymore In our Valley, the so-called subprime mortgage crisis has been more a rumble in the distance than the wolves at…
  2. Media: What’s left for the brand ? A well-established brand is supposed to be a key asset. Everybody keeps dreaming of building a long-lasting brand with lots…
  3. Harbinger increases pressure on the nY Times At least someone believes in newspapers. Harbinger Capital Partners grabbed two seats at The New York Times Company’s board as…

Comments are closed.