So do you see how I could easily be lulled into thinking I shouldn't have to decode strings of text (not binary data!)

Good, cause you must not decode strings of text. Decoding is the process of getting text from "binary data".

The statement $row->{UnicodeCharacter} = decode_utf8($row->{UnicodeCharacter}); is a no-op.

ug. The Unicode bug :(

use strict;
use warnings;

use Test::More tests => 4;

use Encode qw( encode_utf8 decode_utf8 );

my $x1 = chr(0xE9);
my $y1 = encode_utf8($x1);
my $z  = encode_utf8($y1);
my $y2 = decode_utf8($z);
my $x2 = decode_utf8($y2);

is($y2, $y1);
is($x2, $x1);   # Fails

isnt($x1, $y1);
isnt($y1, $z);

1;
[download]

1..4
ok 1
not ok 2
#   Failed test at a.pl line 15.
#          got: 'é'
#     expected: 'é'
ok 3
ok 4
# Looks like you failed 1 test of 4.
[download]

This is better:

use strict;
use warnings;

use Test::More tests => 4;

my $x1 = chr(0xE9);
utf8::encode( my $y1 = $x1 );
utf8::encode( my $z  = $y1 );
utf8::decode( my $y2 = $z  );
utf8::decode( my $x2 = $y2 );

is($y2, $y1);
is($x2, $x1);

isnt($x1, $y1);
isnt($y1, $z);

1;
[download]

1..4
ok 1
ok 2
ok 3
ok 4
[download]