Why your Flow fails silently in production

The flow works in the sandbox, passes UAT, and then quietly stops firing for one profile in production. Here is the checklist that finds it.

The usual suspects

Nine times out of ten it is one of four things: entry conditions referencing a field the running user cannot see, a paused interview waiting on a time path, an element-level fault swallowed without a fault path, or the flow simply not activated after deployment.

Field-level security is invisible to Flow

A flow running in user context silently gets null for fields the user cannot read. No error, no debug line — the decision element just takes the other branch. Run system context deliberately or grant the permission deliberately; never leave it to luck.

Build the alarm before the fire

A one-element subflow on every fault path — log a record, notify a channel — costs ten minutes per flow and turns 'users have been silently affected for three weeks' into 'we knew within the hour'.