DEV Community

Cover image for I Analyzed 1 Million dev.to Articles (2022–2026): Here’s What the Data Reveals
Marina Eremina
Marina Eremina

Posted on

I Analyzed 1 Million dev.to Articles (2022–2026): Here’s What the Data Reveals

From the rise of AI topics to publishing patterns and writing habits.

From December 2022 to January 2026, developers published over one million articles on dev.to.

Developer communities often reflect where the industry is heading.

Some trends you can almost feel intuitively. Tutorials seem less common. AI is everywhere. Some topics fade while others suddenly explode.

But intuition only goes so far — it’s much better to look at the data at scale.

So I analyzed 1 million dev.to articles published between 2022 and 2026 to understand:

  • what developers write about
  • when they publish
  • how topics evolve over time

Some findings confirmed what I expected. Others genuinely surprised me 😊

Here are some of the most interesting insights from 1,000,000+ articles.


The AI Takeover: 3% → 23% in Three Years

Let's start with the elephant in the room 😊

I'm sure everyone would bet that AI-related articles have grown. We see it, we feel it intuitively. But what do the actual numbers say?

Back in December 2022, ChatGPT had just launched. The ai tag appeared in roughly 3% of dev.to articles. At the time, it was still a relatively niche topic, mostly discussed by ML engineers and researchers.

Fast forward to late 2025 - early 2026.

18% to 23% of all articles now mention AI in some form.

In other words: roughly one in five dev.to articles now touches AI.

Here's what that explosion looks like as a share of total articles that has the ai tag:

These aren’t just theoretical discussions anymore. Many posts focus on practical implementation — building with LLMs, RAG pipelines, AI agents and developer tooling.

This isn't just a tag trend — it's a fundamental shift in what developers are building and learning.


The Tag Timeline: ai Surpasses webdev and programming

Here are the 25 most-used tags across all three years, ranked by the percentage of articles that include them:

At first glance, this looks exactly as you might expect. webdev and programming dominate the platform, with ai sitting in third place. But averages hide the real story.

Because the tag rankings weren’t static.

By mid 2025, the ai tag had surpassed both webdev and programming, climbing from #3 to #1. Here’s what that shift looks like over time:

Another intriguing observation: after nearly two years of steady decline, the tutorial tag turned upward around mid 2025.

The data show that tutorials are making a comeback. This is one of the patterns that surprised me most and the one I'm most curious about.


Publishing Volume & Engagement Paradox

One of the first patterns visible in the dataset is the consistent growth in publishing activity on the platform.

The trend is clear: article volume has been steadily increasing, reinforcing dev.to’s role as a major hub for developers sharing knowledge.

But the data also revealed an interesting puzzle.

When I compared the number of articles published per month with total reactions and comments, something unexpected appeared. While the number of articles keeps rising, overall engagement trends slightly downward.

In simple terms: more people are writing. But each individual article is receiving less attention.

So what’s happening? One likely explanation is attention saturation. The number of articles published each month has grown significantly since 2023, but developer attention is not unlimited.

In fact, just in January 2026 dev.to had so many articles published that it would take 38 days of nonstop reading to catch up. Who would accept such a challenge? 😅


Most Articles Take 1–6 Minutes to Read

Another interesting pattern appears when looking at estimated reading time.

Nearly 90% of all articles take 6 minutes or less to read.

Most writers aren't aiming for essays. They're sharing a quick tip, a snippet of code, a thought that fit in one sitting.

The platform runs on short-form energy. Which makes sense - it's how most of us consume content these days anyway.


When Developers Publish: Is Tuesday–Wednesday Really the Peak?

You’ve probably heard it many times — Tuesday and Wednesday are the busiest days for publishing on developer platforms. I wondered would the data confirm this pattern?

It turned out, the differences between Monday through Friday are actually minimal — only about 2% separates the highest and lowest weekday.

Weekends, however, show a clear dip, with Saturday and Sunday about 5% lower than weekdays.

I used to wonder if Tuesday was secretly the best day to publish. Now I'm thinking: just publish when you're ready. The day matters less than the content 🙂


Behind the Numbers: How the Data Was Collected and Processed

All of these insights come from a data pipeline built on Azure.

  • Extract: Python scraper running in Azure Functions pulled data from the dev.to REST API
  • Load: Raw JSON landed in Azure Blob Storage
  • Transform: Databricks with PySpark cleaned, deduplicated and shaped the data (Bronze → Silver → Gold layers)
  • Analyze: Gold tables fed into Power BI for visualization

All orchestrated with Azure Data Factory.

Data Source: Forem REST API
Total articles: 1,000,000
Time span: 3 years
Cloud cost: Surprisingly reasonable (Databricks was the biggest line item)


Big thanks to the dev.to team for creating such a welcoming space where developers can share, learn and grow — and to everyone who writes here: you make this community vibrant and inspiring. It’s amazing to see how the platform brings people together and sparks creativity across the community ❤️

Top comments (13)

Collapse
 
ingosteinke profile image
Ingo Steinke, web developer

My reaction to most posts: either ignore or moderate as low quality and flag to admins as spam. In 2022 DEV had to incentivize users to write more. Now it's time to do the opposite. I'd also recommend everyone add the #ai hashtag to their blocklist, only that doesn't seem to work anymore. Slop detection heuristic might also filter out long titles with dashes and tropes like "here is what I learned" ... if that wouldn't filter false negatives like this interesting post that somehow passed my human spam filter. Thanks for sharing!

Collapse
 
marina_eremina profile image
Marina Eremina

Thanks for commenting and sharing your perspective, totally get it! It’s interesting how platform dynamics shift so quickly. I think your idea of using the ai hashtag sparingly makes a lot of sense, it gave me something to think about 🙂

Collapse
 
webdeveloperhyper profile image
Web Developer Hyper

Wow! Great analysis of DEV Community. You really have strong analytical skills. Thank you for sharing! 😄

Collapse
 
marina_eremina profile image
Marina Eremina

Thanks for your kind words! I’m really glad you found the analysis interesting. DEV has a lot of fascinating data to explore, so it was fun to dig into it and share the insights 😄 One shocking part for me was discovering that we now write such short articles, on average only about a 3-minute reading time. Really did not see that coming!

Collapse
 
webdeveloperhyper profile image
Web Developer Hyper

The shocking part for me is that posts are increasing, but reactions and comments are decreasing. I hope the community becomes more active.

Thread Thread
 
marina_eremina profile image
Marina Eremina

I agree, more engagement would be great! I have to admit I also used to think reactions or comments weren’t that important. Now I realize just showing that the article is useful really matters for the author and encourages more people to share their knowledge. Even a quick comment can make a difference :)

Thread Thread
 
webdeveloperhyper profile image
Web Developer Hyper

I think so too! However, the problem is that it takes time for me to write a comment. I wish I could write more smoothly. 😅

Collapse
 
evanlausier profile image
Evan Lausier

Omg please tell me you anonymized the data….🤣🤣hate to see what mine looks like lol

Collapse
 
marina_eremina profile image
Marina Eremina

Thanks for your comment, that’s actually a very important point!

I definitely won’t analyse or share any data about individual authors or specific posts, that’s completely out of the question. No individual data is exposed, so you’re completely safe 😄

Collapse
 
konark_13 profile image
Konark Sharma

Wow, amazing insights. Awesome article.

Collapse
 
marina_eremina profile image
Marina Eremina

Thank you so much! I'm very happy you liked it 😊

Collapse
 
rubasri_srikanthan profile image
Rubasri Srikanthan

Very insightful post, thanks for putting this together.

Collapse
 
marina_eremina profile image
Marina Eremina

Thanks for reading and commenting! Happy you liked it 😊