If you’ve been with us on our journey over the past few years, you would've known that our growth has been through the roof. Thousands of new articles are published on Hashnode each day by some of the best writers in tech.
We’ve been seeing record engagement and views on articles written on Hashnode, and that’s a testament to the strength of the community of writers and readers here on Hashnode. Here's the growth of the number of articles published on the platform, monthly.
However, with an increase in scale, we’ve realized that our existing personalized feed wasn’t able to keep up with the explosion of content and didn’t do a great job of surfacing the best content for our readers.
So we had to rethink the algorithmic feed from the ground up.
Revamping the personalized feed
We studied how feeds in general worked at social media networks, blogging platforms, and other content platforms — and deeply studied the problem at hand.
Before sprinkling AI to it, we wanted to get the basics right; which is why we decided to take an approach that is easier to implement and solves our problem at hand faster.
From a first principles perspective, we narrowed down on three important parameters that would matter to the reader:
Let me break those down for you.
You want to see relevant posts, that's the core expectation. The revamped algorithm now takes into account a variety of factors like considering your followed tags, authors, publications, subscriptions, as well as your reading interests, to personalize your feed. We'll look to add more signals to this list in the future to serve you content that you truly resonate with.
We don’t want to just serve you posts that match your interests. We want to bring to you content that is gaining traction in the dev community — through likes, comments, bookmarks, and others — and surface these articles in your feed. In the future, we'll add more social elements to this and find you content that is recommended by your network within Hashnode and on other social media platforms.
And the final element in this list is quality. Top authors stand out because of their insights and wisdom gained from real-world experience in the field of tech. These authors take their craft seriously and only publish top content. We now have an added emphasis on a range of factors that include but are not limited to authors who are featured, those who have committed to a custom domain or a pro account, and other factors. We'll continue to look at factors such as whether they have received coveted badges, surpassed milestones, and been contributing their wisdom to the community.
The numbers don't lie
We put the new algorithm through the paces and the results were pretty telling. Over a month-long A/B test involving 1,000 active users, we found that 80% preferred the new feed. That's a good sign we're on the right track.
But numbers only tell part of the story. What truly matters is how these changes are impacting the daily experience of our users. And we're thrilled to say that the feedback has been overwhelmingly positive, especially from some of our most active users. Here's what they have to say:
Platform Engineer: 'The feed right now is a lot more to my taste compared to earlier, and the quality of posts is also up.'
Sr. Software Engineer: 'I'm seeing more relevant posts by tags and more to my liking. Overall feels much more curated and clickable.'
These firsthand accounts reinforce that our efforts are not only improving the numbers but also enhancing the user experience in meaningful ways."
Hurdles solved along the way
But it wasn’t so easy. Along the way, we faced several technical challenges while trying to build out the feed to make it scalable and make it dynamic.
The Road Ahead
There's always room to improve. We'll continue evolving the algorithm after we’ve heard more feedback from writers and readers, and will be exploring AI and machine learning to make things better down the line.
As we navigate the challenges and opportunities of rapid growth, our care for our users remains unchanged. Our recent overhaul of the personalized feed reflects this commitment. We've rebuilt the algorithm from the ground up, focusing on relevance, engagement, and quality as our guiding parameters. The early indicators are promising, with both data and user testimonials supporting the positive impact of these changes. See it in action on the homepage.
So as we keep pushing the envelope, we'd love to hear what you think. After all, Hashnode isn't just a platform; it's a community, and communities thrive on dialogue and feedback. Join our Discord to get involved in co-creating Hashnode with us or leave a comment on this post, and I’ll make sure to reply to you. Let’s continue making Hashnode a daily destination for devs.