The visible cost of manual work is time. The quieter cost is judgment: every manual task is also a stream of small decisions, and decision fatigue research suggests the quality of those calls drops as the day wears on. Studies on physician prescribing found decision quality degrading across a clinical day, which is about as clean a setting as you can get.
Agency and business work isn't clinical, but the shape is the same. By late afternoon, naming conventions get sloppy, QA checks get skimmed, and the reply that goes out is the fast one rather than the right one.
Automation's real return isn't only the hours back. It's removing whole classes of decisions from the day, so the judgment that's left gets spent on work that pays.