TL;DR: Show HN posts that work share three traits: a specific claim about what they built (not a category), a live demo or working product, and a founder who stays in the comments for the first two hours. Posts that fail are usually too vague, too promotional, or abandoned the moment they go up.
We’re submitting CrossMind to Show HN next week. Before we do, I went through fifty recent submissions — what got traction, what got ignored — and pulled out the actual patterns. Here’s what I found.
The title is the whole game
Most Show HN posts die in the title. The HN crowd is good at pattern-matching “this sounds like marketing” and scrolling past.
The formula that consistently works: Show HN: [What you built] that [specific technical capability or unexpected outcome]
Compare these:
“Show HN: AI tool for founders” — ignored. Too broad. Every other post this week is also an AI tool for founders.
“Show HN: I built an agent that maps where your early users are — runs in 40 minutes” — gets clicked. Specific capability. Specific time claim. Implies a real experiment was done.
The specificity does two things: it filters for the right audience (founders who have this exact problem), and it signals that the person built something real, not vague software with an AI wrapper.
One pattern I kept seeing in successful titles: they avoid the word “AI” unless it’s doing specific work. “AI assistant” is noise. “Agent that extracts cookie auth tokens to reduce API costs by 80%” is interesting.
The first comment matters more than the post body
HN readers click on the post, read the title and first comment, and decide whether to engage or move on. The post body is often almost entirely skipped.
The first comment should answer: What did you build, why did you build it, what surprised you while building it?
The “surprised” part is underrated. It signals that something real happened — an assumption was wrong, a technical constraint forced an unexpected decision, a user said something that changed the direction. Authenticity shows up in surprises, not in polished positioning.
Bad first comment: “CrossMind is an AI cofounder that handles marketing and growth autonomously. We’re excited to share it with the HN community!”
Good first comment: “I built this after running 69 cold DMs and getting zero replies. I thought the problem was my copy. It turned out the problem was that I had no idea where my actual users were hanging out. This started as a way to solve that specific thing — a research tool that maps which communities have the highest density of the people I’m trying to reach.”
The bad version sounds like a press release. The good version sounds like a person.
Live demo is non-negotiable for tool submissions
Posts with no working product get one of two responses: “interested, will check back when it launches” (meaning never) or active skepticism about why there’s nothing to click.
The bar isn’t high — it doesn’t have to be polished. A working Loom walkthrough showing a real use case beats a beautiful landing page with a waitlist. HN values “built a real thing” over “positioned a real thing.”
For CrossMind, we’re not running a waitlist anymore. The product is live and the Onboarding runs in 40 minutes — so we can point to a functioning demo. That’s the situation you want before submitting.
If your product isn’t live yet: wait. A Show HN with nothing to try gets buried, and resubmitting doesn’t really work.
Timing: the mechanics
HN votes are time-weighted. Early votes matter more than later votes. Your post effectively lives in “new” for about 30 minutes — if it gets enough upvotes there, it surfaces in the main feed and has a chance at the top.
Optimal submission timing is 9–10 AM US Eastern on a weekday. That’s when the highest-engagement HN readers are online. Late Friday through Sunday is the worst window — lower traffic and your post is competing with weekend leisure browsing, not engaged builders looking for interesting things.
Don’t plan the submission and then leave. You need to be in the comments for the first two hours. Questions that go unanswered in the first hour get skipped by later visitors — the conversation is already cold.
What gets buried: the actual failure modes
After reading through the comments on underperforming posts, the failures cluster into a few types:
“Another X” positioning. If your post reads like a variation on something that already exists, HN readers decide it’s covered and move on. “Another AI copilot” is noise. The specific mechanism that makes it different has to be in the title or the first comment, immediately.
Vague problem statement. “Founders waste too much time on marketing” is not a problem. “Founders who’ve launched to crickets have no systematic way to identify which communities have the highest density of their target users” is a problem. The more specific the problem, the more real it feels.
No evidence anything was learned. Pure product pitches feel promotional. Posts that share something the builder discovered — about users, about a market, about a technical constraint — feel like real content. HN is fundamentally a community that values learning over selling.
Defensive in the comments. The HN crowd will criticize. The submissions that turn into good discussions are the ones where the founder engages with the criticism directly instead of deflecting. “That’s a fair point — here’s why we made that tradeoff” works. “We actually address that in our docs” doesn’t.
What we’re doing for CrossMind’s submission
Slug: “I built an agent that finds your first users — maps communities, surfaces threads, runs in 40 minutes”
First comment will open with the 69 cold DM result (concrete failure), explain what we discovered the actual bottleneck was (community discovery, not copy), and describe specifically what the agent outputs (20+ Reddit threads, 15 Twitter accounts, community map) — not the product category.
We’re submitting on a Tuesday morning. I’ll be in the comments for the first three hours. No waitlist, the product is live, Onboarding produces visible output before payment.
The thesis: a Show HN post works when it reads like a genuine experiment report, not a launch announcement. The builder learned something real, built something specific, and is sharing both.
Whether that holds for our submission, I’ll write about after.
If you’re preparing your own Show HN and want to understand which communities your potential users are actually in before the post goes up, that’s exactly what CrossMind maps during Onboarding. Start here.