On Content Quality & Discovery

Content discovery, quality and personalized experiences

Large online channels invest in optimizing click-through-rates (focus in optimizing the promotion of the content in order to attract user’s attention) and tend to ignore qualitative aspects of the content-driven experience (the actual article may prove to be irrelevant, low quality or not satisfying the promise of its promo-message); in many cases powerful images promoting the content create false expectations possibly leading to disappointment after clicking.

Content ‘timelines’ have also inherent issues: they are optimized for viral effects, CTR and other types of top-level engagement; they tend to ignore the actual quality of the content or specific user satisfaction measurements (how useful and informative a piece of content is; how large the gap is between the promise- that is the title and image of the content- and the actual story served). Even the timelines of professional networking sites are poor, full of posts which are naive, ‘empty’ and serve no purpose rather than providing some instant visibility to the person sharing the content.

While a social network-based content discovery sounds like a good idea, in practice the result is typically noisy and poor as people tend to overshare content, affected, primarily, by the promotion of the content (title and the image used to ‘advertise’ content to its target audience). Speaking as a user, I don’t enjoy the effect of ‘naive content sharing’: people easily sharing content based on instant or no-interaction with it — an attractive image and title proves to be sufficient for users to share.

As the result of the overall online media setup, including the culture and the reflexes of the general audience, the typical online user is exposed to false news, opinions appearing as facts, fact-less stories, opinionated articles, miss-interpreted events and noisy feeds of zero-value posts.

From a content perspective, there are three main classes of online users: the ones who create content, those who primarily consume it and those who primarily share it.

The current setup of the industry practically targets (and serves) the ‘passive content promoters’ who are typically driven by personal and/or professional vanity.

We need new models and mechanisms to serve intelligent, personalized high-quality content experiences. We need to take advantage of the advanced technology available to quantify content quality and introduce it as a core component of the media business.

Content quality

Content quality is difficult to define and quantify due to the multiple dimensions involved — has several aspects such as the intrinsic quality of the story (the amount of information, accuracy, proper classification) along with the experience quality — how well is the story/content packaged and served to the end-users; how well the overall content experience is delivered.

A great story, may written in a poor way (for instance, with no proper structure, as a single large block of text, with no references, no rich media or even with poor language, grammar and syntax). On the other hand we can find great-looking pieces of content with no value, zero-information; or even worse- providing false information.

What we need is an objective, global way to measure and quantify the quality of particular pieces of content. Although this is quite easy in reference to the structural/ packaging aspects of content quality, it is quite complex in reference to the intrinsic quality of the content. This is where we a combination of human assessment (editorial, audience) and advanced Artificial Intelligence systems able to extract the context and perform a number of cross validations. The technology is here; what’s missing is the right strategy to reset the media business towards meaningful, quality-driven content experiences.

Content quality, if properly modeled and quantified, can dramatically improve content-driven experiences online; can trigger a major improvement in serving advanced personalization scenarios; can significantly improve the timeline experience served by social and professional networks out there; at the same time, with the right adjustments in the business model, it can very well serve those high-quality publishers and their audiences; it could re-balance what goes viral and what’s not.
One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.