Hard drives fail gradually more often than suddenly. A head weakens, the surface under it develops sectors the head can no longer read cleanly, and the drive's firmware does what it is built to do: it retries, re-reads, recalibrates and retries again, for as long as it takes. From the desk that is a folder that will not open, a copy that starts fast and dies, a progress bar stuck at 43 per cent. The drive is still returning good data; it is just taking ten times longer to return it, and the time is being spent on the weakest part of the surface.
An external drive adds a bridge to that, and the bridge has its own patience. When the drive takes too long the bridge resets it, the computer sees a disconnect, the drive reconnects, and the copy starts again from the same place. That loop can run all night. On the shingled drives inside most modern portables the firmware's background housekeeping falls behind as well, and a drive that was merely weak becomes a drive that cannot keep up with itself.
What harms a drive in this state is exactly what people do to it: a full scan, a defragment, a disk check, a recovery program working through every sector. Each is thousands of extra reads on the weakest heads. The bench reads the healthy regions once at full speed, then comes back for the weak ones in short passes with its own limits on retries and time, and, where the heads themselves are the problem, replaces them first.