Using data to predict audience behavior 📊

We send out this newsletter every two weeks, talking about how organizations and individuals can better communicate with their audience. Read past issues. In this issue, we’re looking at metrics that indicate what a reader wants from your online content.


A few emails ago, we explored how “chunking” your content into neatly defined sections (bullet points, tables, etc.) can make it easier to read. With that being said-

It would be a mistake to say that all types of content warrant the same presentation, or that visual presentation is all that matters.



A 2018 study by researcher Nir Grinberg looked at how people engaged with online news.


(The paper focuses on news posts, but there is much that we can take away for online content in general.)


It examined how the topic, sentiment, and readability of the text affected reader engagement as opposed to visual and dynamic elements of a page, like images, videos, and page layout.


The study gets rather technical, so here are some key insights:


Your audience behaviour data points

Let’s start off with the obvious data points to measure reader engagement:

  • Click through rates
  • Page views
  • Shares

What these metrics don’t tell you is how the reader interacts with your content once they are on the page. Measuring engagement after they click can enable us to make more informed decisions when creating content.


The study looked at 7.7 million page views on online articles (ranging from Tech, How To, Science, Magazine features, Sports, etc.), measuring the following metrics:

  • Depth – How far the reader scrolled in an article.
  • Relative depth – How much of an article was visible on a user’s screen
  • Engagement – How the read interacts with the page
  • Normalized engagement – Engagement relative to article length
  • Dwell time – Total time spent by a reader on the page.
  • Speed – The speed at which the reader is scrolling through the content

These metrics characterized the following likely reader behaviours:

  • Idle time on page – Periods of activity followed by periods of idle time
  • Scanning – Skimming a page
  • Shallow reading – Not getting far in a post
  • Reading – Reading through the post
  • Long read – Reading additional content such as comments

Ok, so how is all this useful?


Measuring these data points will help you answer questions like:


Does your audience want to skim your content? Are they likely to take their time reading it? Do they want more of it or less?


Findings from the study, for example, indicate:

  • Online sports content is more likely to be skimmed by a reader.
  • A magazine feature is often more extensively read.
  • A How To post has a relatively higher percentage of idle time.

You might find your answers by comparing the type of content you plan to create with these findings. Measuring how readers engage with your own web content (ex. with session tracking software) will give you an even better idea of what mode of presentation to double down on for your audience.


See you next time,