“In the day”, as it were, these were an absolute necessity for streamlining the troubleshooting of system faults. Many motherboard components were replaceable (DMA controllers, interrupt controllers, etc.), and the POST card would allow you to quickly and accurately determine which component was the root of your failure.
Furthermore, you might have had “x” number of service consultations on your daily, and quite a number of miles to drive between each on-site location; so, every moment you saved in making an accurate diagnosis both added to your profit margin on that particular call, and allowed you to get to that next location all the quicker–oftentimes allowing for an extra call or two per day (more revenues for you).
Even today, the intelligent use of POST cards and other diagnostic equipment strikes a firm line of demarcation between the computer professional and the “parts swapper”; and will speed-up fault-mode analysis between the various system components still subject to field replacement.
After a number of years (approaching decades, really) you do get to be adept at fault isolation, and can generally take some shortcuts here and there, without compromising the quality of service one renders; but one should only rely upon “substituting a known-good component for a suspect component” as a method of last resort in attempting to isolate faulty components.
I’ve seen more good components ruined than I’ve ever seen justification for the low-ball practice of parts-swapping; so, I’m of the opinion that the POST card pretty-much pays for itself, in the long-run.
These days, POST cards are remarkably cheap. Just this week, I spent $14.99 (US) to replace a couple of old-model cards which cost in excess of $300.00 (US) to purchase each back in the 1990′s; and the new card integrates the functions of both old cards and is far more versatile than the old ones ever were. $14.99 (US) is cheap enough to install the card into a machine and move the displays to the front panel for permanent use.