Lead Gen Empire: scaling 20 Australian comparison sites with one Claude Code playbook

The premise

Lead Gen Empire is a Boring Ventures-owned network of 20 Australian comparison websites covering verticals like private schools, aged care, accountants, weight-loss clinics, emergency plumbers, IVF clinics, family lawyers, conveyancers and 12 more. As of May 2026 the network covers 23,000+ pages.

It’s run by two operators. Without Claude Code, it would need eight to ten.

The problem this case study answers

Most “scale a network of sites” plays fail because the operational overhead scales linearly with site count. By site #5, the founder is full-time on ops. By site #15, the network’s economics collapse under the weight of:

• Per-site content generation
• Per-site SEO maintenance
• Per-site analytics + reporting
• Per-site compliance reviews
• Per-site bug fixes when something breaks

We needed each operational task to scale logarithmically with site count, not linearly. AI made that possible.

What we built

Layer 1: one shared codebase, twenty configs

All 20 sites live in /leadgen-empire/ (an Astro monorepo). Each site has its own site-config.ts defining vertical, brand colour, FAQs, providers, and other variables. The Astro build reads the active PUBLIC_VERTICAL environment variable and produces that site.

This means: a UI improvement made once ships to all 20. A schema fix made once protects all 20. A new feature shipped once benefits all 20.

Layer 2: Claude Code as the content + maintenance engine

Most editorial work is Claude-driven:

• New articles drafted in Claude Code sessions, reviewed by a human editor (Jenn) before publishing
• Per-vertical FAQs generated and refined
• City-page content templated and populated
• Link audits run via a custom Node script after every build
• Legal-compliance audits run via a script that walks the dist HTML for fake reviews, banned health claims, AggregateRating schema (network-wide kill-switch)

Layer 3: the daily Telegram briefing

Every morning at 7am AEST, a script on our Hetzner VPS:

• Pulls GA4 data from all 20 properties
• Pulls GSC data per property
• Aggregates: top 5 sites by sessions, biggest week-over-week movers, top 10 trending queries network-wide
• Drafts a 200-word briefing in Jenn’s voice (via Claude API)
• Posts to a private Telegram chat with Ollie + Jenn

Read in 90 seconds over coffee. Replaces the half-hour of dashboard-clicking it would otherwise take.

Layer 4: deploys

Every site deploys to Cloudflare Pages from its own git branch. GitHub Actions runs the build (with vertical-specific env vars), Astro compiles, Pagefind indexes, Cloudflare publishes. End-to-end deploy time per site: ~2-3 minutes.

Heavy weeks we hit GitHub Actions free-tier limits, covered separately in our internal notes, but for normal cadence the workflow is free.

What it cost

For a 23,000-page network producing growing organic traffic across 20 verticals: this is laughably cheap.

What it delivered

After the May 2026 content blitz + premium-design rollout:

• 23,000+ indexed pages across 20 properties
• Multiple verticals ranking on page 1 for high-intent queries (private schools, aged care, weight-loss clinics, emergency plumbers)
• Network-wide audit catches 99%+ of broken links before deploy (link audit + legal audit both gating)
• Operator headcount remained at 2

What we’d do differently

Three things, in priority order:

1. Build the legal-compliance audit script earlier. Mid-May 2026 we did a network-wide compliance pass, fake reviews, AggregateRating schema, AHPRA s133 issues for health verticals, MARA Code of Conduct for migration. Should have been built and gating deploys from month one of the network, not retrofitted later.

2. Resist the “ship everything to all 20” temptation early on. First instinct was to roll every improvement to every site immediately. Better to ship to one vertical, watch the metrics for 7-14 days, then promote network-wide.

3. Lean on GA4 properly from day one. We were 6 months in before we had clean per-property GA4 dashboards. The data we lost in those 6 months can’t be recovered. Wire analytics + Search Console before you write your second piece of content.

Why this matters for other AU operators

If you’re thinking about running multiple sites, affiliate networks, geo-targeted SEO plays, multi-brand portfolios, the structural lesson is:

• One codebase, many configs
• One AI playbook, applied across every brand
• Audit scripts as deploy gates, not afterthoughts
• A daily ops digest (Telegram, Slack, email) so you spend reading time on signal, not dashboards

This is the same operational pattern that lets a 2-person team run 20 brands. It works for 2 brands too. It works for 100.

If you’re scaling a small portfolio of Australian sites and want help structuring it, we book audits most weeks for similar projects.