That explanation doesn't make any sense to me. There is only one way to have a return code for a batch file, and that is to cause the ERRORLEVEL to be set to some value. Based on the Backup Exec options I have set, the definition of a "successful" batch file execution is returning an ERRORLEVEL of 0. My batch file does not return 0. Therefore it has failed. I don't know how else Backup Exec could determine that a batch file succeeded anyway, unless it doesn't finish in the allotted time (or crashes, which is unlikely).
Your example is flawed, by the way. Calling "set errorlevel=1" is not the same as having ERRORLEVEL set to 1. All you did is create an environment variable called "errorlevel", which is entirely separate from the implicit variable ERRORLEVEL. (Note that I am not saying that uppercase vs. lowercase matters. I am just doing that for clarity.) Run the test that I did above, i.e., run my batch file and then execute "if errorlevel 1 @echo yes" at the command line. It will echo "yes". Now run your example pre-command batch file (with "set errorlevel=1" in it) and then execute "if errorlevel 1 @echo yes" at the command line. It will not echo anything.
Furthermore, the pre- and post-command batch files are almost certainly launched in separate processes, so there would be no preservation of ERRORLEVEL between them. In any case, Backup Exec just needs to check the ERRORLEVEL as returned by the batch file and then if it isn't 0, fail the backup job. This is they way it is documented, and it doesn't work that way.