Bad monk. Don't ill-advice. The OP uses the fastest way to fetch data. Your change would lose all speed:

$ cat test.pl
use 5.16.2;
use Benchmark qw(:hireswallclock cmpthese);
use DBI;

my $dbh = DBI->connect ();

sub bindc
{
    my $sth = $dbh->prepare ("select * from url");
    $sth->execute;
    my %url;
    $sth->bind_columns (\@url{@{$sth->{NAME_lc}}});
    my $n = 0;
    while ($sth->fetch) { $n++ }
    return $n;
    } # bindc

sub ftchh
{
    my $sth = $dbh->prepare ("select * from url");
    $sth->execute;
    my %url;
    my $n = 0;
    while (my $ref = $sth->fetchrow_hashref) { $n++ }
    return $n;
    } # bindc

say "url has ", bindc (), " records";
say "url has ", ftchh (), " records";

cmpthese (10, {
    bind_columns        => \&bindc,
    fetch_hashref       => \&ftchh,
    });
$ perl test.pl
url has 491826 records
url has 491826 records
              s/iter fetch_hashref  bind_columns
fetch_hashref   1.72            --          -87%
bind_columns   0.224          670%            --
$
[download]

Bad monk. Don't ill-advice. The OP uses the fastest way to fetch data. Your change would lose all speed:

The OP doesn't get any data really really fast :) With that recommendation he'd get data, 87% slower, but still plenty fast enough :)

also %row and $row are two different variables, you're binding to %row, but you're printing from $row