Nobody can tell anything is wrong.

The problem is not one bad decision. The team has lost some of the mechanisms that would normally reveal a bad decision: a check stopped happening, a skill faded, a warning became normal, or past success stopped people asking questions. From inside the system, things can still feel fine.

Asking people how things are going rarely works, because blindness often feels like confidence. Test instead, and look across time.

• Run a few cases where the right answer is already known, with and without the tool, and compare.
• Look back: compare what the team expected with what happened, count how many actions from past reviews were completed, and list the problems that have been accepted as normal.
• Ask about change over time: "What do we see now that would have worried us the first time?" and "What's true today that wasn't when this approach started working?"
• Ask someone new, or someone from outside, what surprised them.

Fixes for blindness build in a regular way to notice: stopping points agreed before the work starts, a scheduled review of what's been accepted as normal, practice without the tool, and a habit of comparing what was expected with what happened.