I published 5 dev.to posts in 24 hours about my MCP server. Here's exactly what each one got.
A test of the "consistency wins on dev.to" advice. Five posts in 24 hours about the same MCP server. Same author. Different angles. Public stats below.
The raw numbers (May 19, 2026, ~24h after the first post)
| # | Published | Title (truncated) | Read | Views | Reactions | Comments |
|---|---|---|---|---|---|---|
| 1 | 2026-05-18 | Shipped: first MCP server for ISO 10012:2026 | 1 min | 0 | 0 | 0 |
| 2 | 2026-05-18 | I Put 7 Products on Gumroad. 4 of Them Had No Files. | 3 min | 0 | 0 | 0 |
| 3 | 2026-05-18 | I Did Not Write New Code This Week. (README rewrite) | 7 min | 11 | 0 | 0 |
| 4 | 2026-05-19 | What an MCP server actually returns: CD-SEM 45 nm | 3 min | 0 | 0 | 0 |
| 5 | 2026-05-19 | sympy.parse_expr will run os.system if you let it | 4 min | 0 | 0 | 0 |
Total: 11 views, 0 reactions, 0 comments. One post (the longest one, posted first) got 11 views. The other four got zero each.
What the data says (and what it doesn't)
Multi-post-per-day dampening is real. Posts #1, #2, #3 went up on the same day with the same author. Only the first (or in this case the longest — posts went up in close succession) got any algorithmic distribution. Posts #2 and #3 received essentially zero algorithmic surface area.
Same pattern on day 2. Posts #4 and #5 on May 19 are both still at zero a few hours after publish. Insufficient time to be sure, but the day-1 pattern strongly suggests dev.to caps the daily algorithm budget per author somewhere around "one post."
The 7-minute post outperformed. Longer reading time correlates with surfacing in this sample. n is too small to be a real signal but the post that got distribution was 7 minutes; the others were 1–4 minutes. Worth testing.
No tag was magic. I tried #showdev, #mcp, #anthropic, #metrology, #indiehackers, #gumroad, #sideprojects, #python, #writing, #security. Combinations of 4 each. None of them produced organic discovery in 24 hours.
What I will change next time
- One post per day, not three. The data above is the experiment, and the experiment says don't do this.
- Promote on Mondays / Tuesdays. Saturday/Sunday posts are a known weak window on dev.to. Posts #1–#3 went up on a Sunday (KST).
-
Use the canonical_url field. All five posts were original-on-dev.to. Cross-posting from Medium with
canonical_urlset still gets you onto dev.to algorithm surface area while consolidating SEO juice on the canonical. I did not do this and probably should have. -
Start a series. The Forem
seriesfield groups related posts. Series posts appear linked at the top of each member, which is essentially free internal-link distribution. I will tag the next batch as a single series.
The honest part
Five posts. Eleven views. Zero reactions. Zero comments. This is what build-in-public looks like for an indie MCP server in a regulated niche on day one, with no audience pre-built. If you are about to ship and you expected a soft landing because "dev.to is fair," recalibrate. The platform rewards consistency over time. Not five-on-one-day.
If you are doing the same thing and want to compare notes, the underlying product is a measurement-uncertainty MCP for ISO/IEC 17025 calibration labs. Repo and live MCP below. I'll publish another stats post in 7 days.
Repo: github.com/kyb8801/measurement-uncertainty-mcp. Live MCP: measurement-uncertainty.mcpize.run. Weekly build log: YB's AI Hustle Weekly.
Top comments (0)