When media companies try to transition to an online business model they run into trouble for all the right reasons: they try to produce content as they normally do but then can't recover their costs of producing it. Online revenues are a fraction of their traditional revenues.

There is a huge gap between what content can earn online and what it costs to produce.

That, in a nutshell, is why the entire media industry is in trouble, and why society is going to hell in a hand basket...

But what if you could figure out how much a piece of online content could earn over its lifetime, and then build your business model around that, from the ground up? That's exactly what Associated Content does, and so do similar approaches from Demand Media, and AOL.

This approach won't make the company any friends among professional journalists because their rates are very low. But it does show the value of online content and the scale of the problems facing established media companies in making their online transition.

Associated Content uses algorithms to determine what content to commission and to calculate its likely earning power. It pays people to produce that content, and also pays extra based on traffic.

Most of its traffic comes from natural search, and revenues come from ads sold by its direct sales division, and Google's AdSense.

This morning I spoke with Patrick Keane, CEO of Associated Content. He is an ex-Googler, he used to work with Tim Armstrong at Google, Mr Armstrong is now CEO of AOL and an investor in Associated Content. Mr Keane also worked at CBS Interactive.

Here are some notes from our conversation:

- We have a different approach to Demand Media and AOL, they are focused on news content, we focus on evergreen content, such as how to potty train a two year old. Less than 10 per cent of our content is news related.

- Our algorithms look at keywords and search terms to determine what content to commission. We look at 30 different attributes.

- We now have 2 million unique pieces of content.

- Online advertising rates have been declining but some of our content gets very high CPMs.

- We aren't interested in creating brands, each piece of content stands alone. Associated Media is an antiseptic brand, I'm skeptical about creating new media brands. But we might do some of that through our guest contributor program.

- We are a private company so we don't disclose revenues but more of our revenue comes from our direct sales than from Google AdSense.

- Brands like our content because many brands don't want to advertise on social networks, they don't know what the content is like. We have a term we call "brand action" that describes the brand association with our content.

- All of our content is checked for plagiarism and we also have humans checking most of our content for quality. But it is impossible to check every piece of content.

- We allow some of our contributors to auto publish because they have built a reputation for quality content.

-We have a different understanding of "quality" we look at how useful the content is. We believe that our content exhibits authenticity.

- We've been asked if we could provide publishers with an OEM version of our platform, but that's not in our plans right now.

- I think that traditional media should adopt a more scientific approach to determining reader interest and demand creation. But it might be too late for some.

- - -

Please see Patrick Keane column in AdWeek: The Case Against Engagement

I recently walked out of a meeting with a potential advertiser who uttered a familiar refrain: "We're looking for engagement." Their stakeholders, they explained, look for them to buy media on sites with strong metrics for "time spent" and "page views per visit," pieces of data that are unrelated to the success of the ads themselves.

We need to stop looking at those engagement metrics as the sole qualifiers of a successful Web site and start looking at how the site fits into the user's overall Web (and consumer) behavior. In an age of granular data collection and analysis, it's time that engagement lost its relevance.

Here is a PearlTree, a visual collection of web pages related to this interview: