Why most automations lose money, and three tests to filter them before you build

The author, who builds automations mainly with n8n, argues that many workflows look cost-saving in demos but generate hidden expenses like tool subscriptions, AI model fees, server costs, and maintenance time. Savings only count if they reduce actual spending rather than just effort, so builders should first name the recurring cost an automation removes. A realistic ledger must include build time, ongoing upkeep, and the cost of failures that go unnoticed for weeks. The piece recommends a 90-day review for each workflow to decide whether it keeps its place. It also advises keeping core components on infrastructure the builder controls, such as self-hosted tools and swappable API keys, to avoid vendor price increases.