Agencies love to talk about results. They don’t love to talk about costs. Here’s the full breakdown of what it actually takes to manage 23 WordPress sites across 10+ industries with a team that’s smaller than you’d think.
The Infrastructure
Five knowledge cluster sites run on a single GCP Compute Engine VM. Monthly cost: under . The other 18 sites are spread across WP Engine, Cloudflare, and client-owned hosting. Our Cloud Run proxy — which routes all WordPress API calls to avoid IP blocking — costs pennies per month because it only runs when called.
The local AI stack — seven autonomous agents running on a laptop via Ollama — costs exactly zero dollars per month in recurring fees. Site monitoring, SEO drift detection, vector indexing, email preprocessing, content generation, news reporting — all local, all free after the initial build.
The Tool Stack
Our total SaaS spend is embarrassingly low for an operation this size. Metricool for social media scheduling. DataForSEO for keyword and ranking data. SpyFu for competitive intelligence. Notion for the command center. Google Workspace for the basics. Claude for the heavy lifting. That’s essentially it.
Everything else is custom-built. The WordPress optimization pipeline. The content intelligence system. The cross-pollination engine. The batch draft creator. These exist as skills and scripts, not subscriptions. Once built, they run indefinitely at zero marginal cost.
Where the Money Actually Goes
The biggest expense isn’t tools or infrastructure — it’s the time required to build and maintain the systems. Every custom pipeline, every skill, every automation represents hours of development. But those hours are an investment, not a recurring cost. The SEO refresh pipeline we built three months ago has processed hundreds of posts since then without any additional investment.
The second biggest expense is content creation itself. Even with AI-assisted generation, every piece of content needs human judgment: is this actually useful? Does it represent the client accurately? Would I put my name on this? The AI accelerates the process dramatically, but it doesn’t replace the editorial function.
The Takeaway
You can run a serious multi-site content operation for less than most agencies spend on a single client’s tool stack. The trick is building systems instead of buying subscriptions. Every hour spent on automation pays dividends across 23 sites. Every process that gets encoded into a reusable pipeline removes a recurring cost from the ledger permanently.
The agencies that survive the next five years won’t be the ones with the biggest tool budgets. They’ll be the ones with the most efficient systems.
Leave a Reply