Wow-Score is a metric for measuring reader engagement

When creating the SEMrush Community, we wanted to be able to understand which of our posts are interesting to our readers and which aren’t. Likes, shares and other social media voting metrics aren’t the best fit for this — people often ‘like’ a post without even reading it. We needed an algorithm for measuring reader engagement, something unbiased, quick and universal for all types of content. This is how we came up with Wow-Score. 2016 release

Goal

To find a discreet way to measure reader engagement.

Why other metrics aren’t as reliable

Understanding whether an article is engaging seems like an easy thing to do. But it only seems that way. Here’s why the conventional metrics fail to measure reader engagement:

Likes and shares are usually a way of showing others that you’ve read something (doesn’t mean you’ve actually read it).
Time on page won’t tell you how much time a visitor spent actively reading (they might have spent half an hour making coffee).
High click rate can be explained by a catchy headline (not high-quality content).
Scroll depth and heat maps sound promising, but in fact they’re quite useless. People tend to read the first few paragraphs of a post more carefully than its rest, which makes heat maps look pretty much the same for most pages.

How does Wow-Score work?

Takes into account scrolling depth and speed.
Takes into account the time of active reading. Knows how to distinguish between active and passive reading.
Takes into account the correlation between text length and how tired a reader is.
Can distinguish between a typical reader and an unusual reading activity.

As a result, we know which articles are more interesting to readers. Each article gets a Wow-Score ranging from 0 to 100.

On the Semrush blog we use Wow-Score to rate our authors:

Results

Wow-Score’s been implemented on the Semrush blog since 2016. It helps our readers find valuable content and assists our editors in discovering the best authors.
The algorithm takes into account numerous factors, but the result is always a single score, which is easy to interpret.
Patent is pending.