Good marketers want to know how audiences will respond to content but the reality is that audiences do not behave consistently. Sometimes a person will skim through the content and other times they will read it closely.  The question is do audiences behave predictively?  There is little question that content behavior and human behavior are complex in general. Even so, we can learn some deeper truths about behavior that are not necessarily complete or exhaustive but accurate and generally robust.Your content strategy guidelines should be built on empirical data that is reproducible by others. You should base your guidelines on data that covers a broad range of situations and has a depth of description.

Nir Grinberg of Northeastern University presented a new study at the 2018 world wide web Conference that provided a large scale empirical analysis of content. He studied a large, client –side log dataset of 7.7 million page views including mobile and non-mobile devices of 66,821 news articles from seven popular news publishers.  The study looked at content from news websites that specialize in different topics including financial news, technology, how to, science, women, sports and magazine features.

The study looked at six analytical metrics that related to Attention:

• Depth-How far a person scrolled in an article (indication of how much content is viewed or read)
• Dwell Time- the total time a user spent on a page (includes things like watching a video)
• Engagement- how much interaction occurred on a page (cursor movements, highlighting)
• Relative depth (how much of an article was visible on a user’s screen)
• Speed (how fast they scroll and an indication of how quickly one “reads” the content)
• Normalized Engagement (engagement relative to article length)

Using a cluster analysis the study identified that those indicators interact to create five content engagement patterns:

• Scanning (skimming an article quickly)
• Reading (reading an article for comprehension)
• Idle (a short period of activity followed by a period of inactivity and then another period of activity)
• Shallow (Not getting far into an article)
• Long read (engaging with the entire article and supplementary materials such as comments)

Through analyzing the six metrics and five engagement patterns the study found that most readers scan sports for scores or highlights rather than reading the full text. Magazines were interesting because they were either scanned quickly at a very high level or read thoroughly with supplementary material read as well. This demonstrates that the reader makes a determination and decision about whether the content looks interesting enough to read in depth.  How To articles had a high percentage of “idle engagements” indicating that readers leave their digital devices to follow the instructions in the physical world.

The study sheds light on how reading characteristics lead to common reading patterns and how distinct reading patterns relate to specific genres of content.  The study also offers more sophisticated ways to measure attention, and indicates that attention patterns vary based on the content genre.

A framework using detailed metrics and patterns can help to provide a baseline of what’s actually happening, and compare it with what might be desirable. For example, what genres of content encourage shallow engagement?  Is shallow engagement ever a good thing? Does shallow engagement provide an opportunity? It is possible people start, then abandon, an article because it is the wrong time for them to view it. Maybe they’d benefit from a “save for later” feature. Maybe the topic is valuable, but the content is uninviting, which brings the engagement to a grinding halt.   With a more sophisticated ability to actually describe content behavior, we are in a better position to consider alternative explanations and scenarios.  

The other issue the study addresses is whether content should conform to typical behavior, or whether content should try to encourage a more efficient behavior.  If How To content involves idle periods, should the content be designed so that people can start and stop reading it easily?  Or should the content be designed so that the viewer knows everything they need to do before they begin (perhaps by watching a video that drills down to how to do the critical steps), so they can complete the task without interruption?   No doubt many people have strong opinions about this issue.  With more precise analytics opinions can become testable hypotheses that can be validated.