Steff
When someone mentions your product on Hacker News, the conversation is usually over within a day. If you see it in time, you can answer questions, fix a misunderstanding or thank a fan while people are still reading. You can catch every mention by searching Hacker News on a schedule and getting the results as a table. This post shows how, and what a month of Show HN launches says about launching there yourself.
What you need
- The names to watch: your product, your company, and two or three competitors.
- A Selextract account. The Hacker News Scraper is ready-made, so there's nothing to build. Runs don't use AI; they run the same tested code every time.
- A spreadsheet, or a place to send the results, such as Slack through the API.
Step 1: Search for your names
Open the scraper and fill in:
- Search terms (optional): your product and company names, up to five
- Search in: Stories and comments
- Posted within: Past week
- Order: Newest first
You get one row per story or comment:
| Column | What it holds |
|---|---|
search_term |
The term that found it |
type |
story or comment |
title |
The story's title; for a comment, the title of the story it's in |
url |
The link the story points to |
hn_url |
The discussion on Hacker News |
text |
The comment, or the text of a text post |
points, num_comments |
Points and comment count, for stories |
created_at |
When it was posted, with time zone |
story_id |
The story a comment belongs to |
Don't skip comments. Many mentions happen in a comment under someone else's story, often when people ask "has anyone tried X?".
Here's what a few rows look like. The product is made up; the columns and formats are the real ones:
type |
title |
text |
points |
num_comments |
|---|---|---|---|---|
| story | Show HN: Fernwood – feature flags in one binary | (empty) | 214 | 87 |
| comment | Ask HN: What do you use for feature flags in 2026? | We moved to Fernwood last year; the audit log alone was worth it. | (empty) | (empty) |
| story | Ask HN: What do you use for feature flags in 2026? | (empty) | 96 | 142 |
Step 2: Check it every day
Your run is saved as a task in your account. Schedule it daily with Posted within set to Past 24 hours. Each morning you get yesterday's mentions, and a thread that's still active is worth a reply.
A few habits help:
- Reply as yourself, and say who you are. Hacker News readers respect a founder who answers questions openly.
- Read the parent story. Open
hn_urlto see what people were talking about before they mentioned you. - Track competitors the same way. When people compare tools in a thread, you learn which features they care about.
Step 3: Study launches before you post yours
Set Search in to Show HN posts, leave the search terms empty, choose Past month and Most popular, and you get the month's best-received launches. That's the best research you can do before writing your own Show HN.
What a month of Show HN looks like
I counted every Show HN post from October 5 to November 4, 2026: 4,247 launches. Here's how they did:
| Points | Posts | Share |
|---|---|---|
| 2 or fewer | 2,907 | 68% |
| 10 or more | 293 | 6.9% |
| 50 or more | 116 | 2.7% |
| 100 or more | 68 | 1.6% |
| 300 or more | 18 | 0.4% |
The median Show HN got 2 points. Only 1 in 60 reached 100.
The topic that shows up most is AI: 1,293 of the 4,247 titles (30%) mention AI, LLMs, agents, GPT, Claude or MCP. But only 13 of them reached 100 points, about 1%. Launches without those words reached 100 points twice as often (55 of 2,954, 1.9%). The month's top launches were small, playful and easy to try: a book list built from Hacker News comments, a city-building game about permits, and a way to talk to your computer without making noise.
My takeaway: a launch competes for attention with 140 others a day. Being one more AI tool doesn't stand out; being something people can try in ten seconds does.
FAQ
Is there a Hacker News API?
Yes, Hacker News has an official API and a public search API. Both return raw items, one request at a time. The scraper handles the searching and paging and gives you a table.
Can I get every comment that mentions my product?
Every comment that Hacker News search finds for your search term, within the period you choose, up to 500 per search term per run. For daily monitoring, that's far more than you'll need.
Why are the points of comments empty?
Hacker News doesn't show points on comments, so there's nothing to collect.
Does it include usernames?
No. You get the text, the links and the numbers.
How is this different from the Who's Hiring scraper?
The Hacker News Who's Hiring Scraper reads the monthly hiring thread and splits each job post into columns. This one searches all of Hacker News. If you're hiring or job hunting, read Every Hacker News "Who Is Hiring?" Job in One Spreadsheet.
Good to know
- Each run takes up to five search terms and returns up to 500 results per term.
- Search matches words, so very short or common names find a lot of noise. Add a second word, like your product's category.
- Runs use your own Selextract credit.