My Top 5 Articles of 2023 — and how much I made writing on Medium

No gatekeeping here.

Photo by Giorgio Trovato on Unsplash

When I first started writing on Medium, my main motivation was to create content that would have been helpful to a past version of myself. As a grad student doing data-focused research, what this meant was explaining technical data science topics that had (finally) clicked for me.

I never expected to get any traction as a writer (I was more of a math and science guy). However, to my surprise, I have gained a (small) audience and generated about $7500 in earnings on this platform.

This is not meant to be a brag (it wouldn’t be very impressive anyway). Rather, as someone who left their (well-paying) corporate job to become a full-time entrepreneur, I want to share how writing on Medium fits into my business and (hopefully) help those considering a similar path.

2023 Recap

Two big things happened in 2023. One, everyone became obsessed with AI. And two, I quit my job. (These are only loosely correlated).

At the beginning of 2023, I worked full-time as a data scientist at Toyota Financial Services, where I’d write data science articles (and make YouTube videos) in the early hours before work and during lunch.

In July, I decided to walk away from that role to go all-in on entrepreneurship. What this meant was making as much content as possible to drive inbound leads for my independent consulting business.

And, to my surprise, it worked.

A big part of this success was the organic reach of my articles on Medium. Here, I’m going to break down my top 5 articles of 2023 and pull back the veil on what (IMO) made them a success.

Top 5 Articles — by earnings

#1: How to Build an LLM from Scratch

My highest-earning article of 2023 (by far) was on how to build an LLM from scratch.

How to Build an LLM from Scratch

• Reads: 19.8k
• Earnings: $1,178.61

This was the final article in a 6-part series on how to use LLMs in practice. The main motivation for this series was that it gave me an excuse to do a deep dive into large language models (LLMs).

I find this approach to writing (and learning) extremely effective because one better understands one's audience when just a few steps ahead of them.

While there are likely many factors that drove the popularity of this article, two factors that stand out to me are as follows. One, it was published in Towards Data Science (TDS), and two, it was boosted by Medium’s curation team.

Each of these factors drives distribution in different ways. The first drives external traffic (i.e. TDS shares it with their networks, and Google SEO tends to like their articles), while the second drives internal traffic (i.e. boosted posts are recommended more on Medium)

#2: I Spent $675.92 Talking to Top Data Scientists on Upwork

The next article reviewed lessons learned from 10 interviews with top data scientists from Upwork.

I Spent $675.92 Talking to Top Data Scientists on Upwork — Here’s what I learned

Reads: 9.2k

Earnings: $497.74

While $675.92 might sound like a lot of money to spend talking to strangers on the internet, it turned out to be one of the best investments I made as an entrepreneur.

The biggest benefit was the relationships I developed with freelancers ahead of me whose mentorship has helped accelerate my consulting business. However, putting that aside, when adding up content earnings and a referral from one of the freelancers, these interviews led to about $1400 in revenue (which is why I did it again).

There are two main things that (I think) made this article stand out. First is the title (it’s got a dollar figure in it and sounds interesting). Second, the article speaks to a specific niche of data scientists interested in freelancing.

#3: Pareto, Power Laws, and Fat Tails

Somewhat unexpectedly, the 3rd article on the list is a more recent one, which kicked off a series on Power Laws and Fat Tails. This goes to show that even in the current AI storm, there is still an appetite for content on the fundamentals of statistics and data science.

Pareto, Power Laws, and Fat Tails

Reads: 1.6k

Earnings: $315.15

The major feedback from this one was the article clearly explained a notoriously misunderstood topic. This is reflected by the number of claps (541) and responses (10) the article generated despite the relatively low read count (1.5k).

Similar to article #1, this article benefited from being published in Towards Data Science and boosted by the Medium curation team, which drove both external and internal traffic.

#4: Fine-Tuning Large Language Models (LLMs)

Next on the top 5 list is the article about fine-tuning LLMs.

Fine-Tuning Large Language Models (LLMs)

Reads: 23k

Earnings: $271.87

Interestingly, this article got more reads than the #1 article on this list but generated a fraction of its earnings. This is (likely) due to the recent change in Medium’s Partners Program, which gives more weight to the number of claps and responses an article receives than reads.

This topic was also a hit on YouTube, where it quickly became my most-watched video (55.5k views) and drew in over 1.7k subscribers (as of writing this).

This is one big benefit of creating on multiple channels. Namely, it makes you more robust against changes from any one platform.

#5: A Practical Introduction to LLMs

The final article on the list kicked off the LLM series, which included #1 and #4 on this list.

A Practical Introduction to LLMs

Reads: 9.3k

Earnings: $235.50

The fact that 3 of the top 5 articles come from the same series points to two things. One, people love LLMs (but we already knew that). And two, series are powerful.

This latter point is due to one driving factor — network effects. In other words, if someone reads an article, it increases the probability that they’ll read another article in the series.

This is compounded by the usual network effects due to referrals, where if someone reads a (good) article, it increases the probability that they’ll refer another reader.

These network effects are what drive viral content. And, inadvertently, this series-based approach is how I’ve made content since the beginning.

Looking Back

At the end of 2022, I had 710 followers on Medium and generated $2,270.19 in earnings for the year. For 2023, those numbers are 4,716 and $5,447.69.

This highlights the power of content creation: it doesn’t scale linearly but exponentially. The catch, however, is exponential growth is painfully slow in the early days (it took me 3 years to get here).

To cope with this pain, I find it is important not to do it for a number (e.g. followers or earnings) but instead for something meaningful to you.

For me, the top reason I make content is to learn. So it doesn't matter whether I have 10 followers or 10,000. Either way, I’ll learn by writing.

Thanks for reading! If you have any questions, feel free to drop them in the comments 😁

Keep up with The Data Entrepreneurs:

The Data Entrepreneurs

👉 Join the Discord | 🎥 Subscribe on YouTube | 📅 Events Calendar

Resources

Connect: My website | Book a call | Ask me anything

Socials: YouTube 🎥 | LinkedIn | Twitter

Support: Buy me a coffee ☕️