Hacking Hacker News

Thirty posts worth of data including optimum title lengths, link destinations, and best day of the week.

Over the last three months, my cofounder and I have used content creation as an exclusive way to generate traffic for our newly launched startup called Whttl. (We named it after the ability to whittle down all on-demand startup options for a given ZIP code and use case.) At the risk of sounding arrogant, it’s worked out well. Quite well. In just 90 days, we managed to drive over $15,000 worth of monthly traffic to our site for next to nothing.

It partly helps that we genuinely enjoy writing about on-demand startups. It also helps that the community we are writing for (early adopters, life hackers, tech aficionados) is genuinely interested in hearing what we have to say about utilizing startups to #winatlife.

I’ve previously written about how to gain traction for posts via a variety of methods, but one avenue is leaps and bounds above the rest — Hacker News. A well received post can send tens of thousands of readers to a site in just a few hours. In one particular blog post of ours entitled Uber vs. Lyft: An Insider’s View, Hacker News referred over 11,000 users in the span of just over a day. It was up voted 116 times, and peaked at #13 on the front page. Since it was such an outlier, I omit it from some of the insights below in an effort to tell a more accurate story.

A screenshot of my Medium stats page. “news.ycombinator.com” is the URL for Hacker News.

There are only a few sites in this world with so much traffic generation potential on tap for the average user. You don’t have to be connected to the media, pull in a favor, or pay for the site visits with Hacker News, commonly abbreviated to “HN”. Instead, the digital democracy that is the HN community validates and promotes your post via “upvotes.”

Although Hacker News gives everyone a fair chance at a top spot, it isn’t always a perfectly fair environment. Some decent posts make it to the top, while mind warping, truly exceptional posts don’t escape the gravitational pull to launch into five-or-more-vote orbit.

So, I began thinking. Which of my posts were doing the best? Were there any meaningful insights I could extract from the data? Did certain days of the week do better than others? What about the length of the title? My curiosity took over, and I dove in.

Since I joined Hacker News nearly two years ago, I have posted exactly 60 times. The bulk of the posts have been submitted after late October of this year. I chose a sample of the most recent 31 posts to pull data from.

Title Length

Perhaps in a future post, I’ll dive deeper into the subjective material of the title to harvest some granular findings. For now, I’m sharing something purely objective — the number of characters (including spaces) in the title of the submission.

This title is exactly 73 characters, including spaces. It received 7 upvotes. See the product we were promoting here.

A definitive pattern emerged. For posts that received 2 votes or less, they had an average of 61 characters in their title. For those who received either 3 or 4 votes, their titles were a bit shorter at 52 characters. Finally, posts that did well, as in 5 or more votes, had average lengths of just 50 characters. But it’s not as simple as just assuming the shorter the better. For the handful of my posts that had 6 or more upvotes, the titles had creeped back up to 51 characters. While I admit this latter point had too small of a data set to rely on, it was interesting to find that the sweet spot was between 50–52 characters.

Upvotes compared to post title length.

For the quartile of my posts with the least amount of characters, they had a mean upvote of 17.86, and a median of 4. (The outlier Uber Vs. Lyft post with 116 upvotes, as previously mentioned, really jacks up the mean. If we omit it, the mean comes down significantly to 3.86.) The quartile with the longest titles had a mean of only 2.63 and a median of just 2.

There is undoubtedly a correlation between title length and post success.

Link Destinations

While I usually submit links that direct to my personal Medium blog, I will also link to my LinkedIn account and Official Whttl Blog. Periodically, I’ll throw in a link from an independent source. (For example, when I guest contribute for Pando, I’ll submit the corresponding URL.)

“Medium” includes the outlier post with 116 upvotes. “Medium (adj.)” excludes it.

It isn’t a far stretch to conclude that users are more likely to click on something when they see it comes from a credible source. I don’t have any posts from top tier sources like New York Times and the Wall Street Journal, but I suspect they generally do well. While I truly love Medium, anyone can post to it, so readers probably discount it as such. (Whether that is conscious or subconscious is unknown.) While we are working diligently to build the brand and credibility of our Whttl blog, it’s still an unknown to most users. Again, users aren’t as apt to jump to an untrusted source. In the end, recognizable outlets seem to do the best.

Days of The Week

It appears that Thursday and Saturdays are the best days to post. I’ve heard for years that Mondays tend to be too busy to grab attention, while Fridays are reserved for daydreams of weekend plans. My data seems in line with that. I must admit that this sample size is still rather small, so as I aforementioned, follow us on Twitter to get notified when we release new posts where n is greater than 100 or so.

Final Thoughts & Reflection

Hacker News still surprises me. Last month, I posted two links a day apart. The content of the articles was exactly the same. Granted, the titles and links were a bit different, but it’s nonetheless fascinating to see the 8X difference between the two. Some unidentified catalyst led to an uncontrollable chain reaction that led to impressive stats for one submission, and a mediocre number for the other.

Hard to conclude what exactly caused #18 to to so much better than #19.

We have no intentions on stopping content generation and promotion anytime soon, so we’ll continue to rack up the data. Perhaps with a sample size of a 100 or more, new insights can be flushed out. Follow Whttl on Twitter, or check out the Whttl blog to stay tuned.

If you found value in this, it would be tremendous if you scrolled down a little further and hit the “Recommend” button.

Greg Muender is the founder of Whttl, described as the “Kayak.com for startups.” Use it to find the startups that have launched in your ZIP code and #winatlife. Drop Greg a line via greg<at>whttl/dot/com. For further reading, check out the Official Whttl Blog.