On Unix systems, utilities return a zero exit-status if what they did went well, if it succeeded without errors etc. If a utility fails, it returns a non-zero exit status.
This way, it's possible to tell a user with the exit-status what when wrong (see e.g. the "EXIT VALUES" or similar section of some manuals, for example the manual of rsync). In short, if the operation succeeded, the exit-status is zero and if it's not zero, it failed and the exit-status may say why it failed.
The test utility, which you use, follow this pattern in that if a comparison succeeded (it's "true"), it returns zero. If the comparison (or whatever operation it carries out) fails (it's "false"), it returns a non-zero exit-status.
The if keyword takes a utility and acts on its exit-status. If the utility returns a zero exit-status, it takes the default branch, otherwise it takes the else branch:
if test 4 -eq 3; then
echo 4 is 3
else
echo 4 is not 3
fi
if can take any utility:
if grep -q 'secret' file.txt; then
echo file.txt contains secret
else
echo file.txt does not contain secret
fi
if mkdir mydir; then
echo created directory mydir
else
echo failed to create directory mydir
fi