Does launching on Show HN still work?

The short answer: yes, for the right kind of launch, but the odds are much worse than they used to be. The typical Show HN launch in 2025 and 2026 got 2 points and no comments. About 1 in 49 reached 100 points, the rough line for a stint on the front page, down from 1 in 16 in 2022. The number of launches that get real attention each year has barely moved; four times as many people are competing for it. The ones that got noticed were mostly technical things people could try or read the code of, and above all, the founder wrote a proper explanation alongside the link: those launches reached 100 points about 4 times as often as a bare link.

I pulled every Show HN post since January 2022, all 110,710 of them, with their points and comments. Here’s what a normal launch looks like now, what the ones that took off did differently, and what I’d do if I were launching tomorrow.

What a normal launch gets

It’s worth knowing what “normal” is before you launch, so a quiet day doesn’t feel like a verdict on your product. Here’s how the 68,921 launches from 2025 and 2026 did:

60.8% of launches got 1 or 2 points, which on Hacker News means almost nobody saw them: every post starts with 1. 75% of those got no comments at all. Only 9.2% got to 10 points, 3.6% to 50 and 2% to 100. If your launch gets 3 points and one comment, you’re in the middle of the pack, not at the bottom.

At the top, launches that reached 100 points got a median of 66 comments. That’s the thing people actually go to Show HN for: dozens of technical people trying your product and telling you what they think, in public, on the same day.

The same attention, split four ways

Somewhere between 600 and 850 launches a year get real traction, and that number has hardly moved. What moved is how many people are trying. So the odds for any one launch fell from about 1 in 16 to about 1 in 61, and the share that got no comments went from 47.8% to 62.3%.

A lot of the flood is AI. In 2022, 5% of Show HN titles mentioned AI, LLMs or agents; in 2026 it’s 35.1%, more than one launch in three. Those launches reach 100 points about half as often as everything else (1.02% against 1.95% this year). Both numbers fell, so everyone’s odds dropped as the queue got longer, but an AI launch is now competing with a crowd of other AI launches in front of readers who have seen a lot of them.

The thing that matters most: write something

When you submit to Show HN you can add text alongside your link. Most people write a line or nothing. The launches that took off mostly didn’t:

A bare link reached 100 points 1.31% of the time. A full write-up, over about 300 words, reached it 5.65% of the time, more than four times as often, and every step in between goes the same way. It holds whatever you link to: GitHub repositories with text reached 100 points 3.46% of the time against 1.38% without, and websites 2.24% against 1.26%.

I don’t think the words themselves are magic. A founder who writes 300 words about why they built something and how it works has usually built something worth 300 words, and is around to answer questions. But that’s the point. Hacker News readers want the story and the tradeoffs from the person who made it, not a landing page. If I were writing one, I’d cover who I am, what problem pushed me to build it, how it works under the hood, what’s still rough, and what kind of feedback I want.

What kind of launch gets noticed

For 2025 and 2026, 68,921 launches, 1,403 of which reached 100 points (2.04%), here’s how often launches reached 100 depending on what the title said. These are plain keyword matches on the title, so they’re rough, and a title can count in more than one row.

Launches that read like engineering do best: something written in Rust, Go, Zig or C reached 100 points 1.6 times as often as average, open source 1.5 times. Launches that read like a product pitch do worst. Titles with “SaaS”, “for teams” or “for startups” reached 100 points 0.51% of the time, about a quarter of the average, and browser extensions 0.77%.

None of that means you should rewrite your SaaS in Rust. It means Hacker News readers reward things they can poke at and learn from, and mostly ignore things that look like they’re trying to sell to them. If your product is a business tool, Show HN is a lottery ticket, not a launch plan. If you do post it, lead with the interesting technical part, not the pitch.

How to write the title

Length matters less than I expected. Short titles do a little better than long ones, but the gap is small, and very long titles are too rare to say anything about. What the topic numbers above suggest is that what the title says matters more than how it’s shaped: say plainly what it is and what’s interesting about it, and leave out the marketing words.

What to link to

Launches pointing at a GitHub repository reached 100 points 2.51% of the time, other websites 1.85%, and text posts with no link 1.21%. Readers like to see the code, and a repository tells them it’s real. If your product isn’t open source, link to something people can use straight away, without signing up, and put the explanation in the text.

When to post

The best hours were 16:00 to 18:00 UTC, which is morning in California and lunchtime in New York: about 2.84% of those launches reached 100 points, against 0.69% at 07:00 UTC. Midnight UTC also looks good (2.87%), but far fewer people post then, so I wouldn’t read much into it. The busiest hour is 15:00 UTC, with 5,512 launches. The day barely matters: Sunday was best at 2.39% and Wednesday worst at 1.78%, too small a gap to hold a launch for.

Should you launch again?

35% of the 58,033 people in this data launched more than once, and repeat launches were 49% of all launches in 2025 and 2026. I followed each person through their posts to see what happens next:

Practice alone doesn’t help much: a later launch does about as well as a first one. What predicts a good launch is a good previous one. People whose last launch got 10 points or more reached 100 next time 4.75% of the time, more than twice the average, probably because they’ve learned what this audience wants, or have built some trust with it.

Posting the same thing again after a quiet launch is allowed on Hacker News, and plenty of people do it, usually about 9 days later. It works a little less well than a fresh launch: 1.57% reached 100 points and 7.6% got to 10, against 9.2% for launches overall. So a second try is worth it, especially if you change what made the first one quiet: the title, the link, or the missing write-up. It isn’t a reset button.

What to expect in the comments

Comments follow points. Below 10 points, the typical launch gets none. Between 10 and 49 points, the typical launch gets 6 comments; between 50 and 99, 28; and past 100, 66. If you get noticed, expect detailed, blunt questions about how it works, what it costs, and what it does that something else doesn’t. Be in the thread for the first few hours to answer them; on Hacker News, a founder who answers well is part of the launch.

What I’d do

Decide if Show HN is the right place. If your product is a technical thing people can try or read the code of, it’s one of the best launches you can get. If it’s a business tool with a sign-up wall, treat it as a bonus, not the plan.

Write the post properly. A few hundred words: who you are, why you built it, how it works, what isn’t done yet, and what you want feedback on. This is the single biggest difference in the data.

Link to something people can use or read right away. A repository, a demo, a page that works without an account.

Post in the US morning, around 16:00 to 18:00 UTC, and keep the afternoon free to answer comments.

If it’s quiet, fix one thing and try again in a week or two. Most launches are quiet. It tells you very little about the product.

What I can’t tell you

Points aren’t customers. A launch can hit the front page and sell nothing, and a quiet launch can find the five people who needed it. All of this is about attention, not outcomes. If customers are what you’re after, I wrote about which startup subreddits are for what and what indie founders actually make.

It’s also correlation. Write-ups and open-source dev tools might do well because readers like them, or because the people who make them are better at writing for this audience. Hacker News has its own ranking rules, including a second-chance pool where moderators re-surface posts they liked, which I can’t see from the outside. “Their first since 2022” means first in this data; some of those people launched before.

How I did it

Every Show HN post from 1 January 2022 to 23 October 2026 came from Hacker News’s public search API, with its points, comment count, author, link and text as they stood that day. Points arrive mostly in a post’s first day, so older posts don’t have an unfair head start. The topic rows are keyword matches on the title: for example, a title counts as AI if it says AI, LLM, GPT, agent, Claude, ChatGPT or MCP. A launch counts as the same link again when the same person posted the same address before; I read a sample of those and they were the same product relaunched, sometimes with a new title. Hours are in UTC, and the 95% ranges come from the counts. This is part of a bigger dataset of 686,751 builder posts, written up in Building got easy. Getting anyone to care didn’t, where you can also get the data.

Cite as: GTM for One (gtmforone.com), “Does launching on Show HN still work? 110,710 launches, 2022 to 2026”, https://www.gtmforone.com/blog/does-launching-on-show-hn-still-work.

If Show HN is a lottery for products like yours, the other way in is the slow one: work out who would actually buy it, and write to them one at a time. That’s the part I built GTM for One to do.