In February 2026 I published the first post about VantLabs on Twitter/X. I announced I was starting a studio and shared a couple of screens. Six months later, I have real data on what works, what doesn't, and why most "build in public" advice is over-simplified.
The metrics that DON'T matter
The first thing I learned was to ignore three numbers everyone chases:
- Likes. A tweet with many likes and few comments generates considerably less real conversation than a tweet with half the likes but with long replies. Likes are noise.
- Followers. Growing your follower count doesn't automatically translate to growing the real community. When I post a direct question, the people who reply tend to be more or less the same — independent of how much the account has grown. Follower count measures ego, not community.
- Impressions. Platforms tell you a post reached tens of thousands of people. The reality is those people saw less than two seconds of the post in their scroll. That's not reach, it's inflated stats.
The metrics that do matter
After a few months, I redefined what I measure. None of this needs special tools — they're metrics you carry as a mental note:
- Conversations that end in email. Every time someone who saw a post takes the trouble to write directly to your inbox to continue the conversation, that's a real event. Likes are easy; writing an email is deliberate.
- People who repeat. How many distinct people have commented on multiple posts over time. That's the "real audience". Everything else is tourism.
- Time between publishing and the first useful reply. When you post something specific enough, someone who knows the topic replies fast. If nobody replies in hours, the post wasn't specific enough to trigger conversation.
Content types that generate real conversation
Patterns I noticed after several months, ordered by how much they invite a reply:
- Technical decisions with explicit alternatives. "I'm deciding between A and B for this thing. Pros and cons. Thoughts?" → generates long replies from people who know.
- Specific failures with detail of what happened. "My cron crashed because of X. Here's the root cause." → people share similar cases and tips.
- Concrete questions with context. "I have this issue with such-and-such provider in production. Anyone seen it?" → replies from people who lived it.
- Reasoned contrarian opinions. Polarize but generate the best debates, as long as the opinion is well argued and not click-bait.
Content types that generate nothing
- Generic announcements. "We're building something exciting 👀" → zero replies, we all know it's noise.
- Motivational threads. "10 lessons from a solo founder" → likes, zero conversation.
- Screenshots without context. A pretty screen with no text explaining what you're solving → silent admiration, not conversation.
- "Milestone" posts. "Hit 1000 followers!" → secondhand embarrassment, mostly.
The mistake I don't make anymore
A few months ago I published a long post about "the mistakes I made when starting". It was genuine, specific, useful. But it was written as performance: each mistake had a lesson, each lesson a quote, everything closed with a moral. It felt like a self-help book.
It worked on likes and almost nothing on replies. People read, felt good, double-tapped, scrolled. Nobody messaged. I learned that content optimized for audience-as-spectators doesn't generate conversation; it generates consumption.
Now when I write I try to leave the content ugly on purpose: no morals, with open questions at the end, with unpolished sentences. It works better.
What's next
Build in public remains, for me, the best strategy for someone building alone. Not for marketing — for feedback. The handful of real conversations that end with people trying what you build are worth more than any paid campaign I could afford. What changed in these months is the idea of what counts as "publishing".
Before I thought publishing well was publishing a lot. Now I think publishing well is publishing specific. One good post every 5 days is better than a generic post every day. And a good post is one that ends with someone writing me in the inbox, not with someone liking it in their scroll.