The 6-hour window: why you need to know when your product appears on HN
Last year I found out a product I was working on had been mentioned in a Hacker News thread three days after it happened.
The thread had 23 comments. Half of them were questions I could have answered. One of them was a competitor's product, naturally submitted by the competitor, who was there on day one while I was not. The thread had moved past the front page by day two. I found it by searching my product name a week later.
That was the incident that made me build a monitor.
The timing problem
A Hacker News thread lives on the front page for 4-8 hours. Once it falls off, the density of readers collapses. The comments that happen on day one are read by thousands of people. The comment you post on day three is read by a handful.
The same dynamic happens on Reddit. The first comment in a thread about your product — positive or negative — shapes how everyone else in that thread perceives it. If you're not there, someone else fills that space.
GitHub is slower but higher-stakes. A bug filed in a third-party repo that lists your tool as broken is forever. "This library doesn't work with X" gets indexed by Google. It persists. A quick response that fixes the issue or explains the workaround changes the search result from "this thing is broken" to "they responded immediately and fixed it."
The pattern: you built something in public, you can't watch everywhere at once, and the 6-hour window where a mention actually matters closes before you find out about it.
Why Google Alerts doesn't solve this
I tried Google Alerts first. Two problems:
- Google doesn't reliably index Hacker News or Reddit. The search algorithm doesn't surface these sites for keyword queries the way a direct search would. I missed mentions entirely.
- Even when Google Alerts catches something, the lag is unpredictable. Sometimes it delivers same-day. Sometimes it delivers three days later. Three days later is not useful for a HN thread.
The sources that matter for developer tools — HN, Reddit, GitHub, Google News — each have their own APIs. HN has the Algolia API (hn.algolia.com). Reddit has a public /search.json endpoint. GitHub has a search API. Google News has an RSS feed. None of these are hard to hit. They just don't exist in a single product at a price point that makes sense for a solo founder.
What actually matters for a developer tool launch
If you're building something technical and you want to know when people talk about it, you need:
- Hacker News — where engineers discover and discuss tools. The highest-signal source.
- Reddit — where engineers complain about tools, ask for alternatives, and share what they're using. Often more candid than HN.
- GitHub — where your users file issues, where other tools reference you in their docs, where the first "does this work with X?" questions appear.
- Google News — for when you get written up in tech press or a blog. Less time-sensitive but useful to know about.
Twitter is conspicuously absent. Twitter's developer API now costs a minimum of $100/month just for basic read access. That's not viable at a $15 price point, and frankly the signal-to-noise ratio for discovering meaningful brand mentions there has always been low.
The spike problem
The other thing I wanted — and couldn't get from Google Alerts — was a spike alert.
If one person mentions my product in a Reddit thread, I don't need to know about it immediately. If three people mention my product across three separate threads in the same four-hour window, that's unusual. Something is happening. I want to know right now, not Sunday in the digest.
This is the difference between ambient monitoring and real-time signal. Most of the time, nothing is happening — the Sunday digest is fine. Occasionally, something breaks out. That's when you need to be there.
How I built it
Brand Monitor runs two Lambdas: a scanner that fires every 4 hours (HN Algolia → Reddit → GitHub → Google News RSS, in that order, checking since the last scan for each source), and a Sunday digest that collects all the pending mentions for the week and sends one email.
The key design decision: everything goes through a fingerprint-and-dedup layer before hitting the pending queue. A mention's fingerprint is a hash of its source and URL. If you've already been told about a mention, you won't be told about it again — even if the scanner runs while you're in the middle of reading your digest.
Spike detection is a rolling counter per keyword per 4-hour window. If 3+ new mentions come in during the same window, an immediate email goes out. This is the one exception to "weekly digest only."
The whole thing is one SAM deploy. DynamoDB for state, EventBridge for scheduling, SES for email. No servers. Under $5/month in AWS costs at moderate scale.
The free tier
One keyword, HN only, weekly digest. Free forever.
This exists because HN is the highest-leverage source and the easiest one to monitor without a rate limit problem. If you're just starting out and want to know when your product appears on HN, you don't need to pay.
The upgrade trigger is self-evident: you get the weekly digest and find out you had three Reddit mentions you didn't know about. Or a HN thread hit 40 points before you found out. Once that happens once, $15/month is an easy yes.
I'm calling it Brand Monitor. Free tier for HN monitoring. Solo tier ($15/mo) adds Reddit, GitHub, and News plus immediate spike alerts. Team tier ($39/mo) adds Slack webhooks and 10 keywords.
— Chester