Data Analysis

My road to 1.5 Million views on Medium

Analyzing my Reads, Views, Claps, Article length, and more.

Benedict Neo

Published in

Nerd For Tech

5 min readMay 26, 2022

PowerBI dashboard of my medium statistics

I’ve been writing on Medium.com since 2019, and as a data scientist, I decided to analyze the stats on my medium articles for two reasons.

First, I want to see what interesting insights I can gain.

Second, because data analysis is fun.

The start of any data analysis work is, of course, the data itself.

To get the data, I decided to utilize a powerful no-code web scraper called Octoparse, which I previously used to scrape Elon Musk’s tweets, Data Science job postings, YouTube comments, and more.

It’s free to use, but if you want access to the premium features, you’re in luck! Octoparse is having a summer sale on June 15, which means huge discounts on the pricing plans, so you’ll only be limited by your imagination (for scraping the web, at least).

Scraping Medium with Octoparse

First things first, tell Octoparse what website you want to scrape.

Then, click on browsing mode so you can sign in.

Google sign-in doesn’t work due to authentication issues, so we’ll have to be creative.

Click on “Sign in with email”.

Then put in the email for your medium account.

You should be getting a one-time login link in a few seconds

Copy that link, paste it into Octoparse, and head to the stats page.

Once you’re on this page, click on auto-detect in the orange pop-up box.

After it’s done, we have the data!

Click on the button to download it.

Of course, the real work comes in cleaning the data, as web scraping can get pretty messy.

After cleaning + manually adding the article reading length, it’s time for some analysis!

Data Analysis with PowerBI

PowerBI is an awesome tool for building cool dashboards, which will be my weapon of choice for this task.

Here are the visualizations I made from the available data.

Enjoy some cool charts and stories behind the data.

Reading time by Publications

I started out writing without any publications.

Then I started submitting them to The Startup and Better Programming in 2019.

But as I’ve shifted my focus to Data Science, most of my articles are on the Towards Data Science and bitgrit Data Science Publication.

Views and reads over time

I did a series of articles on AI and Data Science resources that gained a lot of traction.

My most viewed/read article was Top 20 free Data Science, ML, and AI MOOCs on the Internet, which is the spike you see in January 2020.

COVID hit in 2020, and I lost motivation for writing. Then, in 2021, I started writing for bitgrit, an AI competition and recruiting platform for data scientists and laid the foundations of their data science publication.

Article reading time

Mike Sall, the Head of Data Science of Medium, wrote an article back in 2013 stating the best post length that kept people engaged was 7 minutes, so I do my best to keep my articles around that length.

Looks like I’m doing a pretty good job.

Minutes vs. Reads

Since the optimum post length is 7 minutes, is there an obvious correlation between minutes and read time?

It appears not in my case.

Let’s dig into that.

We have very few reads for a long post (18 minutes). Does this mean that longer posts have lower read times?

There are other factors in play.

First, I posted this on the bitgrit Data Science publication, which at that time had less than 100 followers. So not a lot of people would’ve seen that article.

Second, this article was pretty technical, so it wasn’t for the average reader.

On the other hand, these three data points had pretty long reading times but still racked up many reads.

First, it’s posted on Towards Data Science, which now has over 600k followers. Second, these were resource articles, so no technical knowledge was required.