Lately I’ve been learning some Javascript and thought it’ll be really nice to get out of the course’s projects and mess around with some real-life implementation (that’s not to say that being the 100,00th programmer that writes a rock-paper-scissors game in Javascript isn’t important 😅).
“How many people read a post all the way through?”
The goal of this project is to use Google Analytics’ conversion goal tracking in order to try and estimate how many people read a post all the way through.
The overall plan was to take the estimated reading time feature and compare it to the time the user is actually spending on the page. If the user spends more time on page than the estimated reading time, it's a conversion and we should send an event to Google Analytics 🎉 if not, don't do anything.
Step 1: Calculate the estimated reading time
I was always impressed by how sites like Medium show an estimated reading time on each post. Apparently that it's largely based on several researches done on how many words per minute we read on average [Read more].
You could go ahead and grab all the words in a post, parse them in Javascript and parse the words. Luckily, Carlos Alexanndro Becker wrote a code snippet in Liquid that does the same thing using content | number_of_words we get for free in jekyll. PHEW 😅
{% highlight liquid %} {% raw %} {% assign words = content | number_of_words %} {% if words < 360 %} 1 min read {% else %} {{ words | divided_by:180 }} mins read {% endraw %}
Step 2: Get the user's time on screen
That was my main Javascript and jQuery practice. Whenever you go into a page, I take a timestamp and then on unload take another and do a quick math comparing it to the estimated reading time we grabbed from step 1.
{% highlight javascript %} {% raw %} {% endraw %}
Open issues, things to improve
There are some open issues that can improve this super simple script: From running with it live around two weeks I've seen that if someone forgets a tab open I still send a conversion event even after 40+ minutes. The general assumption going into this exercise is that someone will take a reasonable amount of time to read a post, maybe one way to approach it should be firing the event if the conversion happen within 5 minutes of goal time.
I'm personally a fast reader, so when it comes to long form writing I may finish reading before the estimated reading time, that's also a consideration. In that regard, from a design perspective there is merit to take people who "skim" through the post and send a different conversion for them.
Anyhow, hope you find this useful ☺️