If you want to count case-insensitively, count the number of a canonicalized version, like lc 'targetL' for example, or fc 'targetL'. You will need to check your hash for all keys who have lc $key eq lc 'targetL' then.

But as you don't show us what your data structure is, or how it comes to be, it is very hard to give you more concrete advice.

My best hint is to calculate the data before or while constructing your data structure, or, if this data structure is the result of a database query, to leverage the database functionality to calculate the number of targetL instances instead.

Shame on me I forgot to post the data structure. Here it is:

$VAR1 = [
          {
            'origin' => 'IB',
            'targetL' => 'Ahnenforschung',
            'sourceL' => ''
          },
          {
            'sourceL' => '',
            'targetL' => 'akzent',
            'origin' => 'IB'
          },
          {
            'sourceL' => '',
            'origin' => 'EU',
            'targetL' => 'Akzent'
          },
          {
            'origin' => 'IB',
            'targetL' => 'Akzent',
            'sourceL' => ''
          }
]
[download]

Expecting:

$VAR1 = [
          {
            'count' => 2,
            'origin' => 'IB, EU',
            'targetL' => 'Akzent' 
          },
          {
            'origin' => 'IB',
            'targetL' => 'Ahnenforschung',
            'count' => '1'
          }
]
[download]

My script works fine, but it counts as different entities 'Akzent' and 'akzent' which I would like to count and put togheter. The data structure is constructed in a complex way, so that this manipulation needs to be performed on the datastructure itself, and not before.

Note that 'targetL' in my first record (output) could be both 'Akzent' or 'akzent' (do not care), but I do not want to put all my datastructure to lc, i.e. I want to preserve the original as much as possible

Then do your counting in a case-insensitive approach:

my %count;

$count{ lc $_ }++
    for map { $_->{targetL} } @$results;

my $winner = $count{ "akzent" };

my @matched = grep { lc $_->{targetL} eq $winner } @$results;
[download]

#!/usr/bin/perl

use strict; # https://perlmonks.org/?node_id=11108780
use warnings;
use Data::Dump 'dd';

my @suggestionsTemp = (
  { origin => "IB", sourceL => "", targetL => "Ahnenforschung" },
  { origin => "IB", sourceL => "", targetL => "akzent" },
  { origin => "EU", sourceL => "", targetL => "Akzent" },
  { origin => "IB", sourceL => "", targetL => "Akzent" },
  );

dd \@suggestionsTemp;

my @suggestionsUnsorted;
my %targets;

for my $h (@suggestionsTemp) 
  {
  push @suggestionsUnsorted, ($targets{lc $$h{targetL}}={targetL=>$$h{
+targetL}})
    unless $targets{lc $$h{targetL}};
  $targets{lc $$h{targetL}}{origin}{$$h{origin}}++;
# $targets{lc $$h{targetL}}{count}++;
  }
for ( values %targets )
  {
  $$_{count} = keys %{$$_{origin}};
  $$_{origin} = join ', ', sort keys %{$$_{origin}};
  }

@suggestionsTemp = values %targets;

dd \@suggestionsTemp;
[download]

Outputs:

[
  { origin => "IB", sourceL => "", targetL => "Ahnenforschung" },
  { origin => "IB", sourceL => "", targetL => "akzent" },
  { origin => "EU", sourceL => "", targetL => "Akzent" },
  { origin => "IB", sourceL => "", targetL => "Akzent" },
]
[
  { count => 1, origin => "IB", targetL => "Ahnenforschung" },
  { count => 2, origin => "EU, IB", targetL => "akzent" },
]
[download]

for my $h (@suggestionsTemp) {
    my $targetL = lc $h->{targetL};
    push @suggestionsUnsorted, ($targets{$targetL}={targetL=>$targetL}
+)
        unless $targets{$targetL};
    $targets{$targetL}{origin}{$h->{origin}}++;
    $targets{$targetL}{count}++;
}
$_->{origin} = join ', ', sort keys %{$_->{origin}} for values %target
+s;
[download]

Output:

$VAR1 = [
    {
        'targetL' => 'ahnenforschung',
        'count' => 1,
        'origin' => 'IB'
    },
    {
        'count' => 3,
        'origin' => 'EU, IB',
        'targetL' => 'akzent'
    }
];
[download]