7 min
What owning 200 websites taught me about automation
Five years of running a fleet of revenue-critical properties, and the operating rules that survived it.
- infrastructure
- automation
- leadership
At some point the number of sites stops being a number and becomes a physics problem. You cannot hold two hundred properties in your head, and any process that assumes you can will fail quietly until it fails loudly.
Rules that held up
- If it happens twice, it becomes a pipeline. If it happens twice a week, it becomes a template.
- Infrastructure as code is not about elegance, it is about being able to answer "why is this site different?"
- Performance is a budget, not a project. Fleet-wide Core Web Vitals only improve when regression is blocked, not reported.
- One pipeline shape for every property. Special cases are debt with interest.
- Centralize the analytics early, even if nobody asks for it yet. They will.
The part about people
Leading a distributed squad across two countries taught me that automation is also a communication tool. A pipeline that anyone can read and run removes an entire category of meetings, and it makes the technical contact for stakeholders a role instead of a bottleneck.
The lesson I carry into everything since: leverage is not working faster, it is making the second, tenth and hundredth instance nearly free.