I launched three programmatic SEO sites on April 23. The architecture is Astro 5 + Turso + GitHub Actions running at $25/month. It's now two months in. The organic search picture is what you'd expect from any new domain: essentially nothing. Google hasn't indexed most of the content yet. The pages that are indexed rank somewhere between page 6 and page 12 for their target queries.
What I'm not doing is waiting for that to change.
The bet I'm making — and which I'll evaluate by October 31 — is that cross-channel distribution will deliver more referral traffic during months 1-6 than any realistic organic scenario could. That's a specific claim, with a specific deadline, and two specific conditions under which I'd call it wrong.
The bet, stated as a falsifiable claim
By October 31, 2026, the combined monthly referral traffic from dev.to, Hashnode, Bluesky, and YouTube will exceed organic search traffic for at least two consecutive weeks.
Secondary claim: at least one of those four channels will show up as a statistically meaningful segment in Cloudflare Analytics — not rounding error, but a source I could identify as worth optimizing further.
I'm deliberately not claiming large numbers. The baseline here is a new site where organic is near zero. "Distribution beats waiting" means even modest referral volume wins the comparison. That's a low bar by design, because month 2 of a new programmatic site is not the time for traffic predictions.
I'll publish the actual Cloudflare Analytics breakdown in month 4. I don't know if the numbers will be embarrassing or not, but they'll be real.
Why distribution beats waiting during months 1-6
The foundational issue with depending on organic search during this window is that Google's indexing and trust pipeline for new domains is slow in a documented, predictable way. New sites publishing programmatic content face a sandbox period that realistically runs 3-6 months before meaningful ranking positions appear — Google's own crawling and indexing documentation notes that crawl scheduling depends on a site's established crawl history, which new domains don't have. I've written about why AI-curated directories face a specific version of this challenge from the AI Overviews angle — but even setting that aside, the basic timeline problem applies to any new domain.
Distribution channels don't have a trust delay. Publish an article to dev.to and it's visible within minutes. Post to Bluesky and it's searchable immediately. A YouTube Short goes live within 24 hours. These channels aren't long-term substitutes for organic traffic — they're available now, when organic isn't.
The second factor is the content library accumulation. Articles on dev.to and Hashnode continue receiving traffic months after publication if they hit the "Top Posts" feed for their tags. A YouTube Short from six weeks ago can still surface in recommendation feeds today. Each piece I publish adds to a corpus that compounds over time — and the compounding starts immediately, not after a 3-month indexing delay.
The third factor is audience fit. Two of the three sites (Top AI Tools and Open Alternative To) target developers. An article on dev.to documenting the technical decisions behind those sites is native content for a dev.to audience, not an ad. The shared Claude Haiku batch client article works as distribution because it describes a real engineering decision that developers who've faced similar batch API problems actually want to read about.
The four-channel stack
dev.to publishes first. Every article in this series goes here as the primary destination, and a content quality gate runs before every upload to catch formatting errors, prohibited tags, and length violations. The canonical URL decision — dev.to as primary, Hashnode as cross-post — came from observation that dev.to's feed algorithm surfaces recent articles more aggressively than Hashnode's.
Hashnode receives every article as a cross-post with dev.to set as the canonical source. Hashnode's audience skews toward longer technical writing and is smaller but more engaged with developer-operator topics. The traffic contribution is lower, but backlinks from Hashnode posts have indirect value for the directory sites' authority over time.
Bluesky runs a daily three-post JSONL queue without an external scheduler. Each post links back to one of the three directory sites. The strategy here is less about direct click-through rates and more about maintaining a visible presence in the communities where AI tools, indie games, and open-source alternatives are discussed. I recently tightened the spec so every findindiegame.com post includes a #steam tag and a direct site link — making the posts more consistent as channel-specific distribution units.
YouTube Shorts runs through an analytics-driven script selection routine that biases tomorrow's content toward whatever archetypes performed better yesterday. Each Short ends with a spoken CTA linking to the relevant directory. This channel has the longest feedback loop — Shorts views to click-throughs takes 24-48 hours to appear in YouTube Studio — but it's the only channel with a recommendation algorithm that can surface content to audiences who've never encountered these sites before.
The counterargument I take most seriously
The strongest case against this bet: distribution channel audiences don't convert, and conversion is what matters.
dev.to readers are developers. They read the article, they find it useful or not, and they leave. They're not looking for an AI tools directory when they open a dev.to article about how I built mine. Click-through rates on embedded links in technical articles are low — typically under 2%. I don't have my own data yet (I'll publish it in month 4), but the pattern is consistent in the industry: developer blogging platforms drive awareness and developer credibility, not downstream product conversions.
Bluesky is more uncertain. The active developer community on Bluesky generates real impressions, but converting a post impression to a site visit for a 60-day-old account with limited followers is a long chain. Building a Bluesky audience takes time in the same way organic ranking does — just on a different and arguably less durable curve.
YouTube Shorts could go either way. A Short that surfaces to someone actively researching AI tools is a genuine acquisition channel. A Short that surfaces to a random viewer who clicks once and bounces contributes to traffic numbers without contributing to any durable audience metric. I don't know yet which pattern dominates for this content.
The three-vertical-directories bet is ultimately predicated on organic traffic building over 12 months. If distribution channels don't create retained visitors, I'm running a treadmill — always dependent on the next piece of content to generate new traffic rather than compounding on past content.
What I'm tracking
| Metric | Target by month 4 | Source |
|---|---|---|
| Monthly referral > organic | Two consecutive weeks | Cloudflare Analytics |
| dev.to click-throughs | >50/month per article | dev.to analytics |
| YouTube Short CTR | >3% across last 10 Shorts | YouTube Studio |
| Bluesky link clicks | >200/month aggregate | Bluesky analytics |
These thresholds are guesses. I chose them as the level below which I'd consider a given channel a vanity metric rather than a real distribution mechanism. Whether the targets are correct depends on what I learn from the data, and I'll revise them in month 4 if they're obviously wrong.
One caveat: Cloudflare Analytics underreports referral sources for visitors using privacy-focused browsers that strip referrer headers. Bluesky click-throughs will be underreported specifically because the Bluesky app strips referrers. I'll note this in the month 4 breakdown.
What would change my mind
Two conditions would make me revise the multi-channel approach and reallocate the time I'm spending on distribution publishing.
First: if Cloudflare Analytics shows zero meaningful referral from all four channels combined by September 30, while organic is showing even modest positive movement, I'd shift focus from distribution to content quality. The automation is lightweight — the Claude Code pipeline handles most publish mechanics — but writing, reviewing, and maintaining the publish infrastructure is not zero cost. If the data says "distribution isn't working, content quality is the lever," I'd follow the data.
Second: if Bluesky and YouTube specifically show click-through rates consistently below 0.5% after 90 days, I'd deprioritize both and double down on dev.to and Hashnode exclusively. Text articles on developer platforms age more gracefully than social posts and Shorts, and they don't require keeping a video generation pipeline running.
Neither condition changes the core claim — I still believe distribution beats waiting during months 1-6. But the channel mix would shift.
FAQ
When will you publish the traffic data from this experiment?
Month 4, which is August 2026. Month 2 numbers would be misleading; indexing is still in flux and referral channels haven't had time to mature. Publishing early numbers would create a false baseline.
Isn't this just blogging to drive traffic?
Partly. The articles document technical decisions that are genuinely non-obvious — the canonical URL cross-posting strategy, the Bluesky queue architecture, the analytics-driven video routine. If someone building something similar finds this useful, that's the primary intended outcome. Site traffic is secondary. Whether that distinction holds up in practice is something the month 4 data will tell me.
How much ongoing time does cross-channel publishing take?
Writing each article takes 20-40 minutes of review and editing on my end. The publish pipeline — dev.to + Hashnode upload, Bluesky queue management, YouTube upload — is almost entirely automated and takes near-zero marginal time per piece once the infrastructure is running.
Do the channels compound or decay?
dev.to and Hashnode compound — older articles continue receiving traffic from their tag feeds. Bluesky posts decay quickly after the first 24 hours. YouTube Shorts have an unpredictable curve — most decay fast, but some continue surfacing in recommendations for weeks. That variance is part of why I'm tracking it as a separate metric rather than treating all four channels as equivalent.
Part of an ongoing 6-month experiment running three AI-curated directory sites. The technical claims here are real; this article was AI-assisted.
Top comments (0)