Reading HackerNews comments for buying signals without getting flamed

The HackerNews front page is the most overcrowded sales venue on the internet. Every Show HN gets 200 comments and a flood of cold emails. Every Ask HN about hiring gets 50 recruiters in DMs. But the comments — the long, rambling, 80-deep t

The HackerNews front page is the most overcrowded sales venue on the internet. Every Show HN gets 200 comments and a flood of cold emails. Every Ask HN about hiring gets 50 recruiters in DMs.

But the comments — the long, rambling, 80-deep threads where actual engineers admit what they are actually using — those are the gold mine.

I have been mining them for three years. The hit rate on intent there is roughly 5x what I get from monitoring posts. And almost nobody else is paying attention because it does not look like a list of leads.

The Show HN author wants attention. The third-tier commenter who says "we are evaluating this against Postgres right now" wants a solution.

The front page is theater. The comments are the conversation.

When someone posts a Show HN, they are performing for an audience. The comments below it are where engineers actually compare notes.

I watched a Show HN for a logging product last spring. The post got 340 upvotes and 180 comments. The author got flooded with congrats and 4 cold emails from competitors.

But three layers deep in one comment thread, a senior infra engineer at a 500-person SaaS wrote: "We're three weeks into evaluating this category and our shortlist is currently Better Stack, Axiom, and Highlight. Anyone have strong opinions?"

That comment had 6 replies and 12 upvotes. It was the most valuable single message on that page that day. The OP of the Show HN had no idea this person existed. The other tools in the shortlist did not see the comment because they were busy congratulating the author.

I emailed the engineer through his personal site, referenced the specific comment, and got a 30-minute call. He did not buy from the company I was helping at the time. But the principle was confirmed: the buyer was in the comments, not at the top.

Most people only watch the front page. That is the arbitrage.

The front page of HN is checked by tens of thousands of people every day. The comment threads are checked by the people in the threads, plus a thin layer of lurkers, plus almost nobody else.

Algolia, which powers HN search, indexes every comment. You can search comments by keyword, by author, by date, by points. Almost no salesperson does this.

The comment-level intent monitoring at Shadow Inbox leans hard on this gap. We watch comments for 4 specific patterns, which I will get to. The point is: the playing field is wide open because the conventional wisdom is "watch Show HN and Who's Hiring."

That conventional wisdom is leaving the best stuff on the table.

The 4 comment patterns I look for.

After three years of pattern-matching, I have narrowed the high-intent comment shapes to four.

Pattern one: the active evaluation. "We're currently evaluating X vs Y." Or "Our shortlist is X, Y, Z." This is a buyer mid-decision, with the budget already allocated and the criteria already defined. If you sell anything in that category, this is a no-brainer. The conversion rate on a thoughtful in-thread reply here is something like 15-20% to a real conversation.

Pattern two: the post-mortem. "We tried X for six months and it didn't work because of Y." This is someone who has been burned and is now in the market for an alternative. Trickier to engage with — they do not want a pitch — but a comment that says "interesting, we hit a similar wall, here's what we did" earns you a reply.

Pattern three: the constraint reveal. "Looking for something like X but for Y use case." Someone is asking a tangential question that reveals a specific need. The intent is implied rather than stated. Lower hit rate per comment, but very high if your tool actually fits the constraint.

Pattern four: the comparison request. "Can someone explain why I would use X over Y?" This is fishing for opinions, but the underlying intent is real — they are about to make a decision and they want validation. Drop a 200-word technical comparison in the thread and you become the expert in their head.

The unwritten code of HN replies.

HN is not Reddit. The unwritten rules are different.

You can be opinionated. You can disagree with the consensus. You can be technical to the point of pedantry. All of that earns upvotes if it is correct.

What you cannot do: be vague, be marketing-flavored, be sycophantic, or hide your affiliation.

The single sentence that has earned me the most goodwill on HN is some variant of: "Disclosure: I work on a product in this space, but here's the honest answer to your question regardless of which tool you pick."

That sentence is permission to give a substantive answer. The crowd reads "honest" and they actually mean it. Lie in the next paragraph and you get torn apart.

The in-thread reply is 10x more effective than the cold email.

Here is the math, roughly.

If I see a high-intent comment and I send a cold email referencing the comment, my reply rate is about 12%. Not bad.

If I post a thoughtful in-thread reply that addresses the question, plus a single-line disclosure that I work in the space, the OP almost always replies in-thread. They feel obligated because the comment is public. About 60% of the time, the in-thread reply leads to a DM or an email exchange.

The math is even better than that, because the in-thread reply also earns me upvotes from lurkers who have the same problem. One good comment on an old HN thread has converted into customers six months later via search.

The cold email reaches one person. The in-thread reply reaches the OP plus the long tail of people who land on that thread later. It is just better leverage.

The DM trap is the same on HN as on Reddit.

The DM-only move feels efficient. It is not.

HN does not have native DMs. You email through whatever contact info is in the user's profile, which most users do not check. The cold-email-from-HN move has the worst of both worlds: it feels intrusive to the recipient and it leaves no public artifact.

I will email an HN commenter only after I have had a public exchange in-thread. That exchange is the warming. Without it, I am just another cold sender who scraped a username.

If I cannot reply in-thread because the thread is too old, I let the lead go. The opportunity cost of writing a cold email to someone who does not know me is higher than the upside of one more attempt. There is always another comment.

Speed matters less on HN than on Reddit.

Reddit threads die in 90 minutes. HN threads stay alive for 24-48 hours, sometimes longer if they keep getting upvotes.

This is a small mercy. It means a thoughtful reply at hour 6 still lands. It means I can take 20 minutes to write a good comment instead of dashing one off.

The tradeoff: HN comments age out of the algorithm fast even within a thread. A reply 12 layers deep at hour 10 is invisible to almost everyone. The window for replying directly to the OP and being seen is roughly 4-6 hours from when the OP commented.

I check HN three times a day instead of constantly. Morning, lunch, evening. That cadence catches most of the intent without making me a screen zombie.

The story of the cold reply that turned a six-figure deal.

I will tell you about one specific exchange because it is illustrative.

A senior engineer at a Series C fintech wrote a comment on an unrelated HN post about query performance. Buried in his comment was: "We are at the point where Postgres is starting to choke and we are looking at ClickHouse, Snowflake, and a few others. Open to suggestions."

I was helping a friend at a smaller analytics company. I did not have time to write a perfect reply, so I wrote a direct one: "If your workload is X% reads to Y% writes and your data is mostly time-series, ClickHouse is the obvious pick. Snowflake is overkill at your scale. (Disclosure: I help out at one of the smaller competitors in this space, and I'd still tell you to start with ClickHouse before you talk to any of us.)"

Three replies in-thread. Then an email. Then a call. Then a six-figure annual contract — not with my friend's company, eventually with a third tool that came up in the comment thread, but the relationship made my friend a trusted advisor and led to two later referrals.

The honest reply that names competitors costs you nothing. The marketing-flavored reply that hides the truth costs you the relationship.

What gets you flamed on HN.

I have been flamed maybe a dozen times. The flames clustered into shapes.

Shape one: replying to a question with a question. HN reads this as evasion.

Shape two: an opening line that sounds like a pitch deck. "Great question — at [company] we believe..."

Shape three: any sentence with the word "leverage" or "synergy" or "transform."

Shape four: a comment that is 80% about your product and 20% about the question.

Shape five: replying to a thread you have not read in full. HN regulars notice when your reply contradicts a higher-voted comment without acknowledging it.

The fix for all of these is the same: write the comment as if you are answering a friend in Slack. Drop the marketing voice. Be specific. Cite a number. Concede a tradeoff. Disclose.

The HN comment as long-term SEO.

A comment I wrote in 2023 about a database migration tool still gets engagement in 2026. Someone hits the thread via Google, reads my comment, and emails me. I get 1-3 of these per quarter from old HN comments.

This is the dirty secret of HN: it ranks well. Google indexes the threads aggressively. A high-quality comment with a specific keyword combo will surface in search for years.

Most "growth marketing" plays expect the play to die in a week. HN comments are the rare channel where the play compounds. A comment in a 2023 thread about Postgres connection pooling is still earning me trust today.

This is also why disclosure matters so much: a comment without disclosure that turns out to be from a vendor will be screenshot-mocked on Twitter five years from now. The internet does not forget.

The tools I use to monitor HN comments.

I will be specific about the stack because the readers of this blog actually want to know.

For free: HN Algolia search with saved searches, plus a personal RSS feed via HN Algolia's RSS endpoints. I get email alerts when someone comments on a thread that matches my keywords.

For paid: I use Shadow Inbox now because I got tired of writing the keyword logic by hand and tuning it. The intent classification on top of the keyword match is what I could not replicate in a weekend. But for the first six months of doing this, I used the free Algolia setup. It is not a wall.

The thing the paid tools do that the free ones do not: classify intent. A comment that contains "evaluating Snowflake" can mean someone is buying or someone is writing a blog post about it. The classifier is the part that filters the noise.

The contrarian thing I keep saying about HN.

Stop watching the front page.

Watch the comment threads on yesterday's front page. Watch the second-page posts that are still gathering comments. Watch the niche subthreads on big posts where the actual engineers are talking.

The front page is the pop concert. The comments are the after-party where people actually talk.

The intent there is older, slower, and richer than anything you will get from a Show HN at hour zero. And almost nobody is competing with you for it.

What about the "but HN hates marketers" objection.

This is the one place I will grant the skeptics a real point.

HN does have a strong anti-marketing immune response, and it is not a paranoid fantasy. The crowd has watched a decade of growth-hackers try to game the platform. They are tired.

If you go in with a sequence-templated reply, a tracking link, and a header image, you will be shredded.

But the immune response is not against operators sharing useful information. It is against the marketing voice and the hidden agenda. Drop both, and the platform welcomes you. The HN regulars I respect most are operators at companies who comment publicly, regularly, and honestly about their work. Patrick McKenzie is the canonical example. Thomas Ptacek is another.

The crowd does not hate marketers. It hates marketing.

That is a different problem and the fix is in your tone, not in your channel choice.

● FAQ

Why are HN comments better than HN posts for intent?

Posts are public theater. Comments are real conversations. The buyer who admits 'we are evaluating Snowflake vs Databricks right now' in a comment is showing more intent than ten Show HN authors combined, and they have ten thousand fewer competing replies.

Will I get flamed for replying to a buying-intent comment?

Only if you reply badly. The HN crowd flames marketing-speak, fake humility, and link-bait. They tolerate technical specificity, opinions backed by numbers, and a clear disclosure. Lead with substance, not your product.

Should I email the comment author or reply in-thread?

Reply in-thread first. A public comment with a useful technical answer earns you the right to be in their inbox later. Cold-emailing someone whose username you got from HN, with no thread context, is the move that gets you blocklisted.

How do I find these intent comments at scale?

HN search by relevance plus the Algolia API gets you 80% of the way for free. Set saved searches for terms like 'evaluating', 'looking for', 'we tried' plus your category. After about 10 categories the manual approach breaks down and you need real monitoring.

What's the single biggest mistake people make on HN?

Pretending to be a neutral user when they're the founder. HN users dig. They check submission history, they cross-reference Github, they call it out in-thread. One outed sock-puppet costs you 18 months of reputation. Just disclose.