← All posts

The article tested its own thesis. It won.

32 readers on dev.to, 1 upvote on IH, and 2 visitors from a one-line GitHub PR I'd merged 18 days earlier. The thesis my article argued — outreach doesn't drive signups, asset accumulation does — won when its own distribution data validated it.

Week 3 update — and a strangely recursive lesson.

Last Tuesday I published a 2,622-word piece on dev.to: "I asked ChatGPT for free status page tools. It didn't mention mine." It argued that signups aren't a function of this week's outreach — they're a function of asset accumulation, funnel fixes, and time.

Then I distributed it.

Two channels, two outcomes:

Dev.to (24h):

  • 32 readers
  • avg read time: 23 seconds
  • 0 reactions, 0 comments, 0 bookmarks
  • ✅ canonical blog post indexed by Google in 24 hours

IndieHackers Show IH (9h):

  • 1 upvote (mine)
  • 1 comment (mine)
  • 0 organic engagement

A week later, the deeper data from Vercel told a sharper story:

  • Of those 32 dev.to readers, 7 (22%) clicked through to my canonical blog
  • IndieHackers sent exactly 1 visitor to my site over seven days
  • github.com — the awesome-status-pages list I'd merged 18 days earlier — sent 2 visitors in the same window, including the founder who became my 4th user
  • And the IH post itself? Four days later, one organic peer comment appeared — "the temptation to check if it worked tomorrow is real" — but zero additional clicks. The comment was the asset, not a distribution event.

The reflexive read is "the distribution failed." It would be hard to argue otherwise on engagement metrics.

But here's what bugged me at 6am this morning, scrolling through the dead thread:

The article itself argued that outreach doesn't drive signups in the moment — assets accumulated do, weeks later. So why was I expecting the distribution of that article to perform any differently?

The dev.to article didn't fail. It got indexed. Three weeks from now, when someone Googles "free Statuspage alternatives indie," that page will be ranking. The IH post got 1 upvote. The IH post will sit on my profile forever.

But here's the part I didn't see coming.

Six hours after I posted on IH (and watched it sit at 0 organic engagement), my 4th real user signed up. I checked the referrer expecting to find indiehackers.com — instead I found github.com. Specifically, the awesome-status-pages list — a one-line PR I'd merged eighteen days earlier.

That same list had already brought one signup nine days after merge. Now it brought a second one, eighteen days after merge. Same asset, two signups, two different weeks. Meanwhile the article I was so anxious about distributing — the one with 32 readers and 0 reactions — the article isn't the asset. The PR from three weeks ago is.

The IH post — the thing I'd been anxiously refreshing every fifteen minutes — moved less traffic to my site than a one-line PR I'd merged eighteen days ago and forgotten about.

This is what's uncomfortable: my own article was a perfect dataset for its own thesis. And the thesis won.

What I'm changing in week 4

  1. Stop measuring distribution as engagement events. The metric that matters is "asset present + indexed + discoverable." Both dev.to and IH cleared that bar. The signup, when it came, came from neither.

  2. Stop reading early data as failure. Comparing 9-hour IH numbers to anyone else's 28-upvote post is recency bias. The cumulative metric is "how many founders, over 6 months, will land on something I made and think — that's the tool I want."

  3. Build the next asset, then the next. This newsletter is one. The blog post version of this newsletter is another. The Twitter thread version is a third. One write, four assets. The awesome-list PR took thirty minutes; eighteen days later it's still working. That's what compounding actually looks like at week 3.

The hardest part of this work isn't the writing. It's giving up the dopamine loop where you check refresh-refresh-refresh and look at numbers that were never the right ones.

I'm 4 users in now (was 3 when I started this draft). Still invisible to ChatGPT. Still uncertain whether week 4 will be the inflection or another quiet week. But I'm done re-checking IH at 6am.

What was your most "the data looked bad but the asset was working" moment? Reply via newsletter and tell me — I read every reply.

— Hao

P.S. The longer version (SQL, screenshots, the 49-minute reply from my first user) is here.