iTunes songs / minute

Well, given that Apple kept providing me with the raw data by updating
their main page with iTunes songs sold, what else was I to do but take
timed samples and compute a songs-sold rate! Once I realized this, I
took samples until (almost) the 95 million songs sold mark, and hacked
out a Perl program to digest the data, and here are the results:


2004070119:00:00: songs / minute = 461.1
2004070119:20:00: songs / minute = 450
2004070119:34:00: songs / minute = 312.714285714286
2004070119:35:00: songs / minute = 2182
2004070119:40:00: songs / minute = 459.6
2004070119:45:00: songs / minute = 490.4
2004070120:13:00: songs / minute = 406.142857142857
2004070120:33:00: songs / minute = 504.55
2004070120:40:00: songs / minute = 671
2004070120:45:00: songs / minute = 458.6
2004070120:51:00: songs / minute = 447.166666666667
2004070120:55:00: songs / minute = 572
2004070121:00:00: songs / minute = 555.2
2004070121:12:00: songs / minute = 419.083333333333
2004070121:30:00: songs / minute = 606.5
2004070121:35:00: songs / minute = 567.8
2004070121:45:00: songs / minute = 544.1
2004070121:56:00: songs / minute = 465.090909090909
2004070122:01:00: songs / minute = 534.6
2004070122:41:00: songs / minute = 584.95
2004070122:50:00: songs / minute = 609.666666666667
2004070123:02:00: songs / minute = 475.5
2004070123:10:00: songs / minute = 689
2004070123:15:00: songs / minute = 577
2004070123:26:00: songs / minute = 514.818181818182
2004070123:33:00: songs / minute = 383.285714285714
2004070123:42:00: songs / minute = 547.222222222222
2004070123:47:00: songs / minute = 555.6
2004070123:51:00: songs / minute = 678.25
2004070123:55:00: songs / minute = 604.25
2004070123:59:00: songs / minute = 587.25
2004070200:15:00: songs / minute = 430
2004070200:25:00: songs / minute = 474.2
2004070200:35:00: songs / minute = 442.7
2004070200:45:00: songs / minute = 403.1
2004070200:55:00: songs / minute = 392.1

37 samples, average songs / minute = 542.068671268671

 

So they take in $3,000 per hour.
What do you suppose they pay out in royalties,
hardware, software, accountants' salary, ...... ?

 

Shouldn't about 500/minute produce $30,000 per hour?

 

Yep. But what's a little zero among friends? (see sig)

 

Very good point.

You have to look at the usage patterns. I would
guess that peak "buying" time is roughly 7-9PM
local time. Now, with there being 4 timezones,
peak buying time would be 7 PM eastern to 9 PM
pacific (midnight eastern). I would also guess
that this peak buying time accounted for about
50% of daily sales.

So, assuming that the data collected was during
a peak buying time ---
500 songs/minute = 30,000 songs per hour.
with 5 peak hours, that's 150,000 songs during the
5 peak hours. If that's half of what
get's sold in a day, then that becomes
300,000 songs per day or about
110 million songs per year.

A lot of guesses and "if's" up there.
i.e. if the data was not collected
during a peak buying time, then songs
per day and songs per year computations
above were probably way too low.

How does this stack up to what apple has
reported for songs sold so far?

 

Obviously the rate varies in time, see below.

More like 470-ish songs / minute.

....

Sun Jul 4 10:16:00 2004: 21098 songs in 77 minutes = 274 s/m
Sun Jul 4 12:43:00 2004: 62031 songs in 147 minutes = 421 s/m
Sun Jul 4 19:55:00 2004: 192159 songs in 432 minutes = 444 s/m

58 samples ...
purchase rates (s/m): min = 237, max = 689, average = 473


The data reduction is trivial, the data collection needs to use
something like LWP, and also post results to a web page. Volunteers
welcomed


#!/usr/local/bin/perl -w

use Date::Manip;
use strict;
my( $otime, $osongs, @rates );

print "\n";

while( $_ = <DATA> ) {
last if /^$/;
my( $time, $songs ) = split ' ';
if( not defined $otime ) {
$otime = $time;
$osongs = $songs;
} else {
my $soe = UnixDate($time, '%s');
my $dtime = ( $soe - UnixDate($otime, '%s') ) / 60;
my $dsongs = $songs - $osongs;
$otime = $time;
$osongs = $songs;
push @rates, $dsongs / $dtime;
printf "%s: %7d songs in %5d minutes = %4d s/m\n",
scalar( localtime( $soe ) ), $dsongs, $dtime, $rates[ $#rates ];
}
}

my $samples = scalar @rates;
exit if $samples == 0;

print "\n$samples samples ...\npurchase rates (s/m): ";

my $max = my $sum = 0;
my $min = 2**31 - 1;
foreach ( @rates ) {
$max = $_ if $_ > $max;
$min = $_ if $_ < $min;
$sum += $_;
}
printf "min = %4d, max = %4d, average = %4d\n\n", $min, $max, $sum / $samples;

__DATA__
07/01-18:50 94762714
07/01-19:00 94767325
07/01-19:20 94776325
07/01-19:30 94780703
07/01-19:35 94782885
07/01-19:40 94785183
07/01-19:45 94787635
07/01-20:13 94799007
07/01-20:33 94809098
07/01-20:40 94813795
07/01-20:45 94816088
07/01-20:51 94818771
07/01-20:55 94821059
07/01-21:00 94823835
07/01-21:12 94828864
07/01-21:30 94839781
07/01-21:35 94842620
07/01-21:45 94848061
07/01-21:56 94853177
07/01-22:01 94855850
07/01-22:41 94879248
07/01-22:50 94884735
07/01-23:02 94890441
07/01-23:10 94895953
07/01-23:15 94898838
07/01-23:26 94904501
07/01-23:33 94907184
07/01-23:42 94912109
07/01-23:47 94914887
07/01-23:51 94917600
07/01-23:55 94920017
07/01-23:59 94922366
07/02-00:15 94929246
07/02-00:25 94933988
07/02-00:35 94938415
07/02-00:45 94942446
07/02-00:55 94946367
07/02-19:15 95285660
07/02-20:17 95305414
07/02-20:34 95311131
07/02-21:24 95331201
07/02-21:44 95341706
07/02-22:00 95351737
07/02-22:28 95364766
07/02-23:00 95383039
07/02-23:59 95412958
07/03-01:21 95446122
07/03-12:20 95617759
07/03-13:39 95645806
07/03-14:20 95662372
07/03-15:01 95676682
07/03-15:57 95699366
07/03-16:26 95715140
07/03-17:09 95731970
07/03-22:04 95869013
07/04-08:59 96024789
07/04-10:16 96045887
07/04-12:43 96107918
07/04-19:55 96300077