The question I keep asking myself is why I launched three separate sites instead of one. A single broad "AI Tools and Open Source" directory would be simpler to operate, easier to link-build, and less confusing to explain. Three domains means three separate codebases to deploy, three sets of ETL pipelines to maintain, three Google Search Console properties to watch — and three domains starting from zero link authority simultaneously.
I made that choice deliberately, with a specific belief about how search works in 2026 that I can be wrong about. This post lays out the bet, the real costs, the strongest argument against the strategy, and the conditions under which I'd consolidate.
The bet, stated plainly
By October 31, 2026 — six months after launch — each of the three sites will rank in the top 20 for at least 15 non-branded long-tail queries in its category. The total organic traffic across the three will exceed what a single broad domain would achieve with equivalent content investment over the same period.
I can't prove this yet. Launch was April 23, 2026. Indexing takes months and the Google Search Console is showing essentially nothing at the time I'm writing this — which is expected, not a signal to panic about. What I can do is state the conditions that would falsify the bet, which I'll do at the end.
The three sites:
- Top AI Tools — model and tool entries pulled from HuggingFace, scored by download rank and likes, with AI-generated summaries, pros/cons, and use cases
- Open Alternative To — OSS replacements for commercial SaaS, filtered by GitHub star thresholds and curated with intro text generated by Claude Haiku 4.5
- Find Games Like — indie game discovery sourced from Steam and itch.io, with genre and mood tagging
All three run on the same monorepo: Astro 5 SSG, Turso libSQL, GitHub Actions for ETL and scheduling, Vercel Pro for hosting. The infrastructure is shared. The data sources, query vocabularies, and quality gates are not.
Why vertical depth over horizontal breadth
The way Google's AI Overviews work — and increasingly how LLMs answer queries — is citation-first. A query like "what's a good open-source alternative to Figma" produces an AI-generated answer with 2-4 cited sources. Those sources are not broad aggregators. They're sites that have demonstrated concentrated expertise on a narrow topic over time.
I wrote about this dynamic in more detail when I bet on AI directories versus Google AI Overviews, but the core argument is: topical authority in 2026 is increasingly query-cluster-specific. Google's helpful content guidance frames this as "depth and breadth of expertise in a specific topic area" — which maps directly to the vertical vs horizontal split. A domain that covers AI tools, OSS software, and indie games doesn't clearly own any of those topics in the way a dedicated directory does. Google's trust signals are semantic, not just domain-age-based.
There's also a structural keyword argument. The queries I care about — "best open source alternative to Figma", "find games like Hades", "open-source LLM for on-device summarization" — are long-tail. Long-tail traffic is the foundation of programmatic content at scale. Each site's long-tail cluster is semantically coherent in a way that a single combined domain's cluster cannot be.
The content quality gates reinforce this split. The noindex gate for programmatic pages uses thresholds that are category-specific: OSS alternatives requires MIN_ALTERNATIVES >= 4 and MIN_TOP_STARS >= 1000. The AI tools site uses HuggingFace download rank quartiles. Indie games uses itch.io rating density. A single broad site couldn't apply all three filter sets coherently without the pages fighting each other's quality signals.
The real cost: link equity and authority fragmentation
Here's where I have to be honest about the primary downside.
Link equity doesn't cross domains. A backlink earned by Open Alternative To is worth nothing to Top AI Tools. If someone writes a post linking to an OSS alternatives page, that authority accumulates only on that domain. In a single-site world, every piece of content I ship, every directory page that earns an organic mention, compounds into one domain's authority. With three sites, I'm building three separate pools from zero, simultaneously.
Scale makes this more concrete. I'm publishing roughly 45 articles per month across the shared CI pipeline. If all those articles lived on a single domain, the authority signal from the content would concentrate. Split across three domains, each domain gets roughly a third of the total publishing output. The compounding is slower per domain.
The EEAT transparency pages I added to each site partially address the trust question that strong brand would solve better. Each site has an explicit "how we rank entries, what the editorial standards are, why this data source" page. But transparency pages don't replace inbound link authority.
Internal linking is also constrained. Astro slug page depth depends on tight semantic relationships between pages. "Open-source Figma alternatives" linking to "open-source Photoshop alternatives" is natural and semantically valid. Linking from an OSS alternatives page to an AI tools page is a stretch that probably carries less topical weight. Cross-domain links are even weaker signaling.
I'm not dismissing these costs. I'm betting they're smaller than the benefit of concentrated topical authority per domain.
Comparison: single domain vs three vertical domains
| Factor | Single broad domain | Three vertical domains |
|---|---|---|
| Link equity | Compounds across all content | Fragmented — three pools start at zero |
| Topical authority | Hard to own any one category | Each site can claim a niche |
| Content management | One codebase, one CI pipeline | Shared monorepo, three deploys |
| Brand clarity | One name to remember | Three names, harder to promote |
| Long-tail keyword focus | Mixed query clusters compete | Semantically coherent per domain |
| Internal linking | Works anywhere across all content | Cross-domain links carry less weight |
| AdSense / affiliate | One site to optimize | Three separate configurations |
| Quality gate tuning | One set of thresholds | Per-site thresholds, more precise |
The "single broad domain" column wins on brand, link equity speed, and operational simplicity. I'm betting the "vertical domains" column wins on the factors that search rewards in 2026: topical depth, relevance concentration, and coherent user intent matching.
The counterargument I take most seriously
The strongest objection isn't about link equity. It's about discoverability through distribution.
A single site with a clear, memorable name is dramatically easier to pitch on Hacker News, Bluesky, and Reddit. "I built a curated directory of open-source alternatives to popular SaaS tools" is a clean pitch. "I built three directories, one of which covers open-source alternatives, one covers AI models, one covers indie games" — that's a worse pitch even if the underlying product is equivalent.
Brand-driven direct traffic — newsletter subscribers, bookmarks, return visitors — compounds differently from search traffic. It requires one thing to remember. I'm sacrificing that compounding for topical search relevance. That's a deliberate tradeoff, not an oversight.
The counter to my counterargument is that I'm not trying to build viral Hacker News posts. I'm trying to build programmatic search properties that earn affiliate revenue at month 12. The audience acquisition strategy is search-first, not community-first. If that assumption changes — if I decide community-driven growth is the right path — the three-site strategy becomes a liability and consolidation becomes the obvious move.
The pairwise model compare pages illustrate this tension: they generate comparison content that serves search queries specifically, not casual browsing. The whole content architecture is optimized for query intent over virality. That's consistent with a three-site vertical strategy and inconsistent with a one-site brand strategy.
What would change my mind
By October 31, 2026, I'll check these specific conditions:
Condition 1: All three sites below 100 monthly organic sessions. Six months of indexing time with consistent content publishing should produce some signal. Zero signal at month 6 would mean either the topical authority hypothesis failed, or the programmatic content approach itself failed. I'd need to distinguish which.
Condition 2: A competitor launches one broad AI directory and ranks for queries I'm targeting in under 4 months. This would mean existing domain authority is more important than topical specificity for this category, which would flip the strategy case entirely.
Condition 3: One domain earns links while the other two don't. Concentrated success on a single site would suggest the winning domain has intrinsic properties I should consolidate onto rather than continuing to distribute effort across three.
Condition 4: The operational cost exceeds 10 hours per week. Three ETL pipelines, three deployment configs, three quality gate parameter sets — if the complexity forces me to spend more than 10 hours per week on infrastructure rather than content, I'm losing the leverage that was supposed to justify the split.
The metrics I'll use: Google Search Console by property, referral traffic in Vercel Analytics, and position tracking for 20 seed keywords per site. I'll publish those numbers once I have 30 days worth sharing. I'm not going to fabricate a current traffic number to make this post feel more grounded.
FAQ
Q: Why not use subdomains instead of separate domains?
Subdomains don't reliably inherit root domain topical authority for separate content categories. ai-tools.example.com and oss-alternatives.example.com on the same root would signal to crawlers that these are related, which partially defeats the vertical specificity argument. Separate domains enforce separate trust signals, separate entity recognition, and separate quality assessments. There's also the honest reason: separate domains force better discipline about content quality and page depth per site, since each site sinks or swims independently.
Q: Does the shared monorepo undermine the separate-domain argument?
No, from Google's perspective the monorepo is invisible. The shared codebase affects operational efficiency — I maintain one Astro config, one Tailwind setup, one GitHub Actions CI template — but each domain serves unique content from unique data sources with unique URL structures. The shared infrastructure is an implementation detail; the content signal is domain-specific.
Q: Can you cross-link between the three sites to build shared authority?
I can, and I do occasionally when there's genuine editorial relevance (linking from an AI tools page to an OSS alternative to that tool, for example). But cross-domain links are both harder to make feel natural and carry uncertain SEO weight compared to same-domain internal links. I don't treat cross-site links as a link-building strategy. The content quality gate checks for forced link patterns — that guard applies to cross-site links too.
Q: If this fails, what's the consolidation path?
Redirect all three domains to one, port the Turso databases into a single schema, rebuild the Astro pages with merged navigation. The monorepo structure makes this less painful than a three-separate-repo setup would be — the three apps already share components and a single build pipeline. Migration would take roughly a week of engineering time. The harder problem is URL redirects preserving any link equity that accumulated on the losing domains. But if the sites have failed to earn meaningful links, that's also not a painful problem.
Related: Why I'm abandoning AdSense on two sites and betting on affiliate monetization and how I built a three-tier content quality ladder for programmatic directory ETL.
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)