Managing one website is straightforward. Managing five gets interesting. Managing forty-three across four business entities, multiple tech stacks, and different markets? That requires a system.
I run METIS Digital, a digital agency that operates a portfolio of web properties spanning real estate diagnostics, professional training, senior services, AI comparison tools, online calculators, and sustainability compliance. Each site has its own audience, monetization model, and technical requirements. Here is how we keep it all running without losing our minds.
Why 43 Sites?
The obvious question: why not just build one big site? The answer is strategic.
Each site targets a specific niche with laser-focused SEO. A domain like "greenwashing-checker.com" immediately communicates its purpose to both users and search engines. A generic "metis-digital.click/tools/greenwashing" does not carry the same weight.
The multi-site approach also provides risk isolation. If one site gets hit by a Google algorithm update, the rest of the portfolio is unaffected. If a particular niche declines, our revenue is diversified across many others.
And frankly, exact-match domains still carry SEO value in certain niches — particularly for comparison sites, calculators, and tools where user intent is crystal clear.
The trade-off is operational complexity. Which is where the interesting engineering challenges live.
The Architecture
Our portfolio runs on three main stacks:
WordPress — about 60% of sites. Used for content-heavy sites where we need flexible publishing, plugin ecosystems, and easy content management. Our diagnostic immobilier sites, training platforms, and content portals run on WordPress.
Pure PHP — about 30% of sites. Used for tool-focused sites (calculators, comparators, checkers) where we need maximum control over performance and do not need a CMS. These sites are lightweight, fast, and cheap to host.
Cloudflare Pages — about 10% of sites. Used for static or semi-static sites where we want edge deployment and zero server management.
All 43 sites are hosted on a single o2switch shared hosting plan (for the WordPress and PHP sites), with Cloudflare in front of everything for CDN, SSL, and basic WAF protection. Total hosting cost: under 10 euros per month.
The Central Nervous System: Supabase
The key to managing this many sites is centralized data. We use Supabase (managed PostgreSQL) as our single source of truth for everything:
- Site registry — all 43 sites with their domains, tech stack, current status, deployment phase, and configuration
- Content pipeline — articles, keywords, semantic clusters, and publishing schedules
- Performance tracking — keyword rankings, traffic data, and revenue metrics
- Health monitoring — automated checks for uptime, SSL expiration, and SEO health
- Credential management — API keys, app passwords, and service accounts (encrypted)
Every tool, script, and automation workflow in our stack reads from and writes to Supabase. There is no spreadsheet, no local file, no "I think it was on the other computer" situation. One database, one truth.
Automation: n8n as the Backbone
Manual management of 43 sites would require a team of 10. Instead, we run a small team augmented by heavy automation.
n8n (self-hosted workflow automation) is our automation backbone. Here are some of the workflows running daily:
Health Monitor — runs every 6 hours, checks all 40 active sites for:
- HTTP status (is the site returning 200?)
- SSL certificate validity
- DNS resolution
- Response time
- Critical page availability
If anything fails, we get an immediate alert. Sites that were fine yesterday but return 500 errors today get flagged before any user reports it.
Content Pipeline — automated workflows that:
- Identify keyword opportunities from Google Search Console data
- Prioritize content creation based on search volume and competition
- Draft articles using AI (Claude API), which then go into a human review queue
- Publish approved content to the correct WordPress site via REST API
- Submit new URLs to Google for indexing
SEO Monitoring — daily checks that:
- Pull ranking data from DataForSEO
- Compare positions against previous periods
- Identify quick wins (keywords ranking 11-20 with decent volume)
- Flag significant ranking drops for investigation
The Deployment Pipeline
Deploying changes across 43 sites requires discipline. Our process:
- Snapshot — before any change, we capture the current state (file backup, database export for WordPress sites)
- Deploy — changes go out via WordPress REST API (for content and snippets) or FTP (for PHP files)
- Verify — automated health check confirms the site still returns 200 OK
- Rollback — if verification fails, the snapshot is restored immediately
We never deploy to all sites simultaneously. Changes roll out in tiers: first to a test site, then to a small batch, then to the full portfolio. This limits blast radius when something goes wrong — and something always eventually goes wrong.
SEO at Scale
Running 43 sites gives us interesting SEO insights that you do not get from managing one or two:
Internal linking across sites. Where it makes sense editorially, our sites link to each other. A calculator site might link to a comparison site for deeper analysis. A training platform might link to a diagnostic service site for practical context. This is not a link network — the links are genuinely useful to users — but the cross-pollination of authority helps.
Content velocity matters. Google rewards consistent publishing. Across 43 sites, we publish 100+ articles per month. Each individual site might only get 2-3 articles, but the portfolio-level velocity means we are constantly feeding the indexing pipeline.
Technical SEO is multiplicative. A Core Web Vitals improvement that saves 200ms per page load, applied across 43 sites, is a portfolio-wide performance upgrade. We invest in reusable optimizations: shared CSS frameworks, optimized image pipelines, and standardized structured data templates.
Keyword research scales. Understanding one niche deeply helps with adjacent niches. Our AI comparison sites share keyword research patterns. Our calculator sites share user intent models. Knowledge compounds across the portfolio.
Cost Structure
People are surprised by how cheaply this runs:
| Item | Monthly Cost |
|---|---|
| o2switch hosting (all sites) | ~8 euros |
| Cloudflare (37 zones, free plan) | 0 euros |
| Supabase (Pro plan) | ~25 USD |
| n8n Cloud | ~20 euros |
| DataForSEO API | ~50 USD |
| Domain renewals (amortized) | ~50 euros |
| AI APIs (Claude, OpenAI) | ~40 USD |
| Total | ~200 euros/month |
That is under 5 euros per site per month for hosting, monitoring, automation, and AI-powered content assistance. The marginal cost of adding a new site to the portfolio is essentially zero — just the domain registration.
Mistakes We Made
Lest this sound too smooth, here are things that went wrong:
Over-engineering early. We spent weeks building a custom deployment system before realizing that WordPress REST API + a simple backup script covered 90% of our needs. Start simple.
Ignoring monitoring initially. We lost a site to an SSL expiration because nobody noticed the certificate had not auto-renewed. Now everything is monitored automatically.
Too many plugins. Early WordPress sites had 20+ plugins each. Performance suffered, updates broke things, and security surface area was enormous. We now run a minimal plugin set and use Code Snippets for custom functionality.
Not centralizing data sooner. For the first year, site information lived in spreadsheets, local files, and peoples heads. Migrating to Supabase as a single source of truth was the single biggest operational improvement we made.
What We Would Do Differently
If starting from scratch today:
- Supabase from day one. Central data from the start, not retrofitted.
- Fewer WordPress sites. More PHP-pure for tool sites, reserving WordPress for content-heavy sites only.
- Automated testing earlier. We now have health checks, but automated content quality checks and SEO validation would have caught issues sooner.
- Standardized templates. We built too many one-off designs early on. A small library of proven templates would have saved months.
The Takeaway
Managing a large portfolio of web properties is less about technical brilliance and more about operational discipline. Centralize your data. Automate your monitoring. Standardize your deployments. And invest in the boring infrastructure that keeps everything running reliably.
The exciting part is not any single site — it is the system that makes all of them work together.
METIS Digital operates 43 web properties across multiple verticals. Learn more at metis-digital.click.

Top comments (0)