Macworld Keynote numbers
Just returned from the show. Apple came in right about where I'd expected. My interpretation - the number ipods (and thus itunes customers) continues to grow rapidly.
- They grew the raw rate of sales back up to the rate from last holiday; 3.8M per day over the period, (they said 5M per day at the peak). this supports the omments of many about seasonality in the market nicely.
- they also slightly picked up the sales per ipod; 1.65 songs per month per ipod, up from 1.32 per month per ipod in the last period, but nowhere near the long term average of more than 3.3 per ipod per month.
17 Comments:
Is "sales per ipod" the most useless statistic on the Web?
And what about the rest of the keynote? Impressions?(although you probably can't say what you really want to say because of the status with your current employer)
Maybe your average (3.3) is completely WONKED since you have them averaging 6 Tunes per iPod on the day of launch!!! (How do you have a MONTHLY avergae data point within the first 24 hours?)
Your interpretation of the data seems sound to me, but I wonder (as the first commenter phil did), is "sales per ipod" really a useless statistic?
There are some issues which cannot be calculated by "sales per iPod".
1. International sales. Yes, we've international iPod stores now, but what about the millions of iPods bought in countries without iTunes Music Stores?
2. iPod upgrades? One could assume more and more people are replacing their old iPods (okay, I concede there is no statistic here) - clearly the number of sales per ipod will fall.
These type of statistics are useful for sites like Paul Thurrott's but that's about it.
These type of statistics are useful for sites like Paul Thurrott's but that's about it.
These statistics are also useful when you're working on the Zune and trying to justify the pathetic sales associated with that device.
Someone's been reading this book.
David, they didn't say 5 million per day at peak. They said they are doing that on a regular basis now.
It was less than 319 days to 1 billion which is an average of almost 3.2 million per day. I would say the holiday period average much more than 3.8 million (remember we have Sept. and October numbers in this "reporting" period).
mcx, there is another flaw you should consider: the size of the catalog. Yes, Apple's offering is much larger with 3.5+ million tracks, but this is a fraction of the true offerings of music available. That's another constraint.
But the most fundamental flaw in this statistic is that it's predicated on the absurd notion that anyone with an iPod must consume 100% digitial downloads. It doesn't take into account that I could still be buying 100% of the avergae consumer CD purchaser's CDs AND say 15% of the average digital consumer's downloads.
Maybe even more flawed is the notion that this presents a problem. I see this stat (with allits flaws) as a sign of health: this stat equates to about 18-23% of CD sales. And that's assuming that 100% of all iPod owners are buying 100% digital downloads only... in a market that the music studios were afraid to enter 2.5 years ago, one they are still resistant to expand. (Oh no!)
Can you stop using this lame 3.3 "average"of an average over the span of the Store. 5 of your largest data points are bunched up in two 2 and 3 month intervals, but your 2 lowest data points span a 6 month interval and a ONE YEAR interval making an average of these "monthly" averages nonsensical.
I can except some value to this mode of analysis (although my own analysis actually differs from yours) but comparing 1.32 to 3.3 makes no sense at all.
David, what # of iPods are you using since they didn't accounce ANY?
Are you projecting 85 million based on 65 + analysts estimates of 19-21 million units for the quarter? extrapolating from past data?
Funny how you claimed you were subtracting from the IB, but when I use the full installed base of 85 million (projected but probably on target since Jobs is probably waiting to say 100 million at an upcoming event) and do the same calculation, I get 1.65 songs per month. How can I possibly get the same number if you are attempting to account or a smaller IB and I am not?
yeah. this time I was absurdly cavalier (on the road) and used 60M as the number of ipods. Probably conservative.
Just checked my spreadhseet and I actually used 70M (not 60M) as the placeholder number for average ipods during the period.
Mae Culpa!
Yes, on reflection, "songs per iPod" is pointless when done in this way because it omits, or slides past, lots of important data.
1) When the first (US) iTunes online music store opened, there were lots of people who had iPods but zero songs from the iTS. (Obviously. Nowhere to buy them from.) So they bought lots of songs in a rush - hey, it's new! This distorts the number of songs per iPod at the start. You'd expect that the rush won't last, and will settle down.
2) When other stores open in other countries, you get the same rush effect. Then it settles down.
3) Some countries don't have an iTunes Store, but do sell iPods (as has been pointed out). The more this happens, the more the "songs per iPod" won't be representative of anything.
4) The more models of iPod there are, the more purchasers will include people who don't have broadband, so are unlikely to buy a song. That pushes down the "songs per iPod" too.
5) we don't know how many people buy multiple iPods, which will also lower the apparent "songs per iPod".
What this all means: the number of songs sold does not, and shouldn't be used to, measure anything about iPods, and vice-versa. It's a statistic that needs to be handled with lots of caveats.
"Just checked my spreadhseet and I actually used 70M (not 60M) as the placeholder number for average ipods during the period."
Then your numbers are wonky again. I get 1.65 using 85 million (they were at 65 over 3 months ago and this holiday they sold more than 5 million....) If you used 70 and did your .84*IB correction with a final adjusted IB of 58.8 your average should have been failry higher. Would you be willing to share your data, and I'll share mine? You are definitely doing something wrong, but I'd be happy to have you scrutinize my data and calculations as well.
average sold during the period would be higher. I just threw 70M in as an estimate of average number of users for the new datapoint. Might be low or high.
"average sold during the period would be higher. I just threw 70M in as an estimate of average number of users for the new datapoint. Might be low or high. "
"average" what? iPods? No sh!t... this ain't no Zune (well 5 million would be huge for Zune, but...) I am still asking you to validate your numbers because they are wrong!!
500 million over 119 days is 127.8 million per month (119/30.4167 = 3.912 months ... 500/3.912 = 127.8). 3.912 per month / 70 million iPods = 1.826.
And this is without adjusting userbase which you claimed to be doing.
This number is not 1.65.
Now you say you are just pluggin in placeholders. I'm happy to debate the merit of these numbers, but let's make sure we have accurate numbers first. So far... by what you yourself have told me, yours are wrong. Let's fix that. Particularly if you intend to continue to use these numbers.
David,
I have some serious issues and doubts with your latest numbers. When you get a chance, take a look and see what you think of my numbers.
-Palmer
The reason it's so silly to use songs per iPod figures as a measure of how successful the iTunes Store is this:
If 10 Zunes were sold and 50 songs were sold from the ZuneStore, that would make the ZuneStore hugely more successful than the iTunes Store.
By this method, the smaller the user base, the easier it is to make the small guy look better.
As far as CD sales go, please also bear in mind that when anyone buys a CD they're automatically buying about 10 songs, some of which they don't really care about. So CD sales will always be "artificially" higher than iTunes sales with its pick-and-choose method. However, I personally think song downloading will not really, really take off whoever sells them, until DRM is removed. It's the DRM that's holding people back, not the sound quality, not the method, not the price, in my view.
palmer, great stuff. My own analysis is in accord with yours. David really needs to explain himself. I want to confirm my own numbers (I've done so many diffferent analyses and averages and have been trying to determine what's wrong with David so I want to review my formulas and data entry for any issues), but I'd like to share my data with you and see what's up. It would be nice if David would do the same because his are pretty apparently wrong... at least the first and last data points.
Post a Comment
<< Home