Success of iTunes Music Sales
This is in response to the "Nerd Fight" regarding the success of iTunes Music Sales.
Before I begin, I am an investor by night and intelligence analyst by day. I'm sure David Caulton knows more about the concerned industry than I do, but I can't find any data to positively back his claims. Most importantly, only Apple knows the true sales numbers and the way they trend day to day and month to month. Apple has said that Forrester's report is "simply incorrect." Even though the iTunes Store is (directly) inconsequential to profits, would Apple risk more shareholder lawsuit by lying about iTunes Store's sales and growth?
My main point here is to show that Apple's public numbers releases only provide enough data to create a very general trend curve and cannot reliably demonstrate David's assertions of mid-reporting period trends.I am unable to find resources for David's claim of a 3.6M song per day period. David, pointed out that I missed some iTunes milestones surrounding last holiday season. We will have a better idea of the trends on Jan 9 during Steve Jobs keynote and on Jan 17 when Apple releases holiday quarter results.
The history: This all started from a Forrester research press release based on 2700 iTunes credit and debit cards transactions over 27 months. That would be a .004% sample of iPod owners based solely on credit card data. But that's not the case, it was 2700 credit card transactions. They also mention that, "iTunes Revenue Is Dominated By A Large Number Of Small Transactions." I've had over 100 transactions in the past 27 months. Therefore, 2700 transactions divided by "a large number of small transactions" could mean the total number of people sampled was 27 or 100 or 2000. Considering that at least two transactions are required to show a per customer trend, the sample is now less than .002%. Is two transactions a trend? Three? Four? A .001% sample? Do the transactions show that the person owns an iPod or if they just wanted to burn one mix-CD? I don't know and I'm not paying for the report, which again, Apple says is "simply incorrect." Continuing...
The report was then picked up by the media which erroneously drew a conclusion of a 65% decline in iTunes music sales. Josh Bernoff of Forrester (and the original report) posted a blog shortly after the media attention that read, "Tunes sales are NOT plummeting! Press credibility, on the other hand..." Then tech pundit Rob Enderle voiced his support for Forrester with "Apple's 'Catastrophic' Decline: Apple's Collapse" without ever acknowledging Forrester's rebuttal of their own story's wake. Oh, Enderle is an ex-Forrester researcher with close personal ties to Microsoft. Next Paul Thurrott of SuperSite for Windows, WinInfo Daily News, and Windows IT ProMagazine blogged to show support for Enderle, huh?
Why are none of these known Apple critics acknowledging that Forrester said "Our credit card transaction data shows a real drop between the January post-holiday peak and the rest of the year, but with the number of transactions we counted it's simply not possible to draw this conclusion... as we pointed out in the report. But that point was just too subtle to get into these articles." Too subtle to get into the articles? Any reporter or analyst worth a dime could conclude that a sample of less than .002% is inconclusive. Bloomberg quietly altered their reporting after Comscore joined in with a NON-Apple supported study which claims that "iTunes music sales rose 84 percent in the first nine months of 2006 from a year earlier." So what's the real story? Only Apple knows.
The odd part for me is that iTunes average song downloads per day increased 10% in the 200 day period from Feb 23 to Sept 12, 2006 over the previous period (averaged 220 days across Christmas) which included the post Christmas 2005 download frenzy. Sales rates increased 10% for the period these folks are claiming they dropped. Of course, we can't see the trend during that period. As Apple entered this period sixty days after Christmas, were sales still riding the post-Christmas wave only to decline? Or had the post-Christmas wave already crashed with sales now on a steady rise with new iPod owners?
David Caulton, recently joined the debate about iTunes Store's success. David says he has used only Apple's public information to obtain his findings. I find a few faults with his methods.
First, David believes that the success of the iTunes Store should depend on iPod sales. Actually, if anything it should be measured against iPod owners since David says his research shows that 15% of iPods belong to owners of multiple iPods. There is an obvious connection between the iPod and iTunes, yet there is no reason one could not be successful without the other. There is not a single reason that songs have to be purchased from iTunes, it's simply an option among dozens of methods for obtaining music. Furthermore, I know over a dozen people who are/were buying music from iTunes, sans iPod. I also know dozens of people with iPods who have never bought a song from iTunes Store - many buy an iPod just to hold their entire CD collection. The device and the store can exist independently. For example, Napster, Yahoo and the now shuttered Microsoft Music store all operated without a required device. And, Creative, iRiver and many others all sell devices without a requirement for an online music store. iTunes Store and iPod only compliment each other, they are not dependent on each other. Only Apple knows how many songs each individual buys, their patterns and whether they claim to own an iPod or two or more. Still, David has linked the iTunes sales to iPod sales to show a downward trend.
Second, David has pointed out that from Oct 25, 2005 to Jan 10, 2006 Apple sold 3.29M tunes per day, from Jan 10 to Feb 7, 2006 Apple sold 3.7M tunes per day and from Feb 7 to Feb 23, 2006 Apple sold 3.125M tunes. All these periods are affected by the post Christmas iTunes download bonanza. What kind of songs-per-day records does iTunes set on Christmas Day, 5M? 10M? Likely higher. Based on the the trend toward Christmas 2005 (ending at Oct 25), iTunes was selling around 1M songs per day. But then from Oct 25, 2005 to Jan 10, 2006 Apple sold 3.29M songs per day. Most recently, sales have settled to 2.5M per day which includes the after Christmas addition of 14M iPods. We can use a graph to approximate that iTunes was selling less than 1.5M per day up until Christmas. Therefore, within 15 days a 60-day running average of 1.5M songs per day was increased to 3.29M. That requires extremely high numbers on Christmas. iTunes reached such large numbers this Christmas that iTunes was hit or miss for most of the day. That extremely high Christmas Day 2005 number quickly fell to 3.7M two weeks after Christmas and 3.125M four weeks later. With such a large increase in sales directly related to Christmas Day, can we get anything more than a general trend from these numbers?
iTunes song milestones are listed at the end.
Third, I have used the same data to create a graph that very much resembles David's graph of iTunes Songs per iPod. My graph has slight variations and for very good reason. Data is only available in periods as long as 220 days. Obtaining sales curves cannot be done with any degree of accuracy. Furthermore, the dates of iPod sales are released every quarter, which cover a period of ~90 days. Finally, none of the figures have been released on the same dates. So now we have an approximation of 220 days of iTunes sales to an approximation of 90 days of iPod sales. Of those numbers, the reported periods are offset by weeks. I think most people can see the large room for error here. First, iPods could accumulate through the quarter like any of the following graphs.
How do you decide which one to use to calculate with iTunes sales? Then there are the holiday sales which likely trend as so,
I realize you're probably thinking, according to this whole debate's theory, that the iTunes sales should be declining as we approach Christmas day. While this could be true, there's also the fact that from Feb 24, 2006 to Sept 12, 2006 iTunes was selling an average of 2.5 Million songs per day; the highest non-Christmas average daily sales in its history. This was a full 60-days after the post-Christmas song rush.
Next, is the problem with averaging. Because the song sales are averaged over longer periods than the iPod sales, the iPods receive an increase much more often. This means that positive trends of iPod sales are calculated with older and lower averages of songs sales. As with my results, David's sales change direction wildly at the point of public sales announcements. The true graph is likely much more rounded, never reaching the highs and lows that the graph represents. Because of averaging and offset dates, songs per iPod usually wane toward the approaching public song milestone release. They then immediately increase as new higher song sales averages are compared with older iPod averages until the next iPod data is released. There is also the case of the huge peak in iPod sales leading up to Christmas followed by a huge spike in iTunes song sales immediately following Christmas. Again, there are too many unknowns to accurately build the graph David presents.
Next, we'll look at the sales data from a slightly different perspective.
This graph shows us a few things. One, iPod and iTunes sales have trended nicely. Both continue to demonstrate growth, although most recently growth appears be slowing. The most recent data indicates that iTunes music sales seem to be growing slightly faster than iPod sales. This will change during the holiday quarter. Now focus on the red line indicating the total number of iTunes songs sold per iPod. The dots indicate when song sales data was released. Notice the slow down or decrease in songs/iPod each holiday period, highlighted in yellow. This is due to sudden surge in iPod sales during the holiday quarter. Then notice the slow increase of songs to iPods as the next holiday season approaches. Guess what, it probably doesn't really look like that.
This graph doesn't represent the situation any better than David's graph. To create this graph we simply plotted the numbers that Apple provided. The problem still lies in the fact that the numbers are released over too long of periods and not released at the same time. For example, July 17, 2005 to Feb 23, 2006 represents one period of iTunes song sales yet it stretches through three quarters of iPod sales. It also begins and ends inside those quarters. The true graph likely shows a slower song to iPod slope all the way to Thanksgiving day. Followed by a drastic decrease up to Christmas day. Then a very quick rise in songs to iPod sales immediately following Christmas. Apple admits to overwhelming Christmas day iTunes sales.
Because we are only plotting points and not trying to average down to a monthly or worse, daily basis, we can see that ending Sept 30, 2006, Apple had sold 67M iPods and as of Sept 12, 2006 they sold 1.5B songs. This allows us to approximate that there are an average of 22 songs per iPod. There are likely more since we can add 18 more days of songs sales. At an average of 2.5M per day (we know the dangers of averaging this long of a period), and counting David's theory that 15% of iPods are owned by the same person we can come to a final tally of 1.545B songs and 57.5M iPod owners for an average of almost 27 songs per iPod owner. But, since we don't know how many people buy CDs to fill their iPods and how many buy online music, we can't conclude anything about buying trends. Looking back at the graph, since the iTunes milestone for Holiday period 2004 was released immediately before Christmas, it is wildly affected by the holiday quarter iPod sales. Then since, the next quarter includes holiday iTunes sales and post-holiday quarter iPod sales it is also wildly affected. Accounting for these inconvenient offsets, it is likely that Apple has maintained a steady 22 songs per iPod ratio or a 27 songs per owner ratio since late 2004. Still, only Apple knows the way this data really looks...
Finally, let's look at my best approximation of the iTunes sales trend. There is no data to show a decline in sales. I have averaged across the holiday 2005 period. All the sales are still there; only the spike has been removed. I'm not looking for day-to-day or even month-to-month trends. I want to see what the iTunes store is doing. For this we see a clear decline in growth (leveling-off) which mirrors the slowing growth of iPods, but there is not enough data to prove a decline in sales. We will have to wait for future data to determine if there is a true decline in sales. Once a decline is identified, only a major change to the iTunes store would likely change it - perhaps a universal DRm or DRM-free songs. Apple and the music industry will have to discover what exactly is holding people back.
I suspect we will continue to see the same trend for at least another year. That is, a slow but steady rise up to Christmas Day then a sharp rise followed by another slow steady climb to Christmas Day. I don't expect to see a sudden mid-year upswing unless Apple or the RIAA can change the rules. To maintain growth, Apple needs to sell 500 million songs in under 200 days. The clock started on Sept 13, 2006. The cut-off date is March 31, 2007. With holiday sales to be included in those numbers, I expect to see the aforementioned sharp rise. And guess what, that won't answer any questions.
More recent posts indicate that David is more concerned with the music industry's situation. He's concerned that digital sales will never make up for the decrease in physical sales. Did we ever assume it would? CDs force people to buy songs in lumps of 10-20 songs. How many CDs have you bought for one, two or if you're lucky 3 songs? Sure there are many CDs worth owning the whole album. However, most of my collection was purchased for the few songs I wanted. When I got married, my wife was in the same boat. As I ripped her CDs into iTunes she put sticky notes on each CD that read, "Tracks 1,3 and 9" or "Only Track 5." Now that I can buy single-serving songs how often will I buy the CD? I admit I became fans of several bands after hearing their other tracks on CD. I may lose out on experiencing some great bands with single track downloads. I think piracy picked up speed because of its single-serving nature. Oh, and it's free. So with people buying 2-3 songs at a time instead of ~ 12, will digital ever make up for physical losses? No. I used to pirate a lot of music. Two songs weren't worth $14.99 and there was no other legal option. I've replaced everything I pirated and more...
Now I can sleep at night...
Data Used:
04/28/2003
Store Opens
05/15/2003
1 Million/18 days
52631 songs/day
06/23/2003
5 Million/39 days
102564 songs/day
09/8/2003
10 Million/76 days
65789 songs/day
12/15/2003
25 Million/107 days
140186 songs/day
03/15/2004
50 Million/90 days
277777 songs/day
07/12/2004
100 Million/128 days
390625 songs/day
12/16/2004
200 Million/156 days
641025 songs/day
07/17/2005
500 Million/212 days
1415094 songs/day
02/23/2006
1 Billion/220 days
2272727 songs/day
09/12/2006
1.5 Billion/200 days
2500000 songs/day
Before I begin, I am an investor by night and intelligence analyst by day. I'm sure David Caulton knows more about the concerned industry than I do, but I can't find any data to positively back his claims. Most importantly, only Apple knows the true sales numbers and the way they trend day to day and month to month. Apple has said that Forrester's report is "simply incorrect." Even though the iTunes Store is (directly) inconsequential to profits, would Apple risk more shareholder lawsuit by lying about iTunes Store's sales and growth?
My main point here is to show that Apple's public numbers releases only provide enough data to create a very general trend curve and cannot reliably demonstrate David's assertions of mid-reporting period trends.
The history: This all started from a Forrester research press release based on 2700 iTunes credit and debit cards transactions over 27 months. That would be a .004% sample of iPod owners based solely on credit card data. But that's not the case, it was 2700 credit card transactions. They also mention that, "iTunes Revenue Is Dominated By A Large Number Of Small Transactions." I've had over 100 transactions in the past 27 months. Therefore, 2700 transactions divided by "a large number of small transactions" could mean the total number of people sampled was 27 or 100 or 2000. Considering that at least two transactions are required to show a per customer trend, the sample is now less than .002%. Is two transactions a trend? Three? Four? A .001% sample? Do the transactions show that the person owns an iPod or if they just wanted to burn one mix-CD? I don't know and I'm not paying for the report, which again, Apple says is "simply incorrect." Continuing...
The report was then picked up by the media which erroneously drew a conclusion of a 65% decline in iTunes music sales. Josh Bernoff of Forrester (and the original report) posted a blog shortly after the media attention that read, "Tunes sales are NOT plummeting! Press credibility, on the other hand..." Then tech pundit Rob Enderle voiced his support for Forrester with "Apple's 'Catastrophic' Decline: Apple's Collapse" without ever acknowledging Forrester's rebuttal of their own story's wake. Oh, Enderle is an ex-Forrester researcher with close personal ties to Microsoft. Next Paul Thurrott of SuperSite for Windows, WinInfo Daily News, and Windows IT ProMagazine blogged to show support for Enderle, huh?
Why are none of these known Apple critics acknowledging that Forrester said "Our credit card transaction data shows a real drop between the January post-holiday peak and the rest of the year, but with the number of transactions we counted it's simply not possible to draw this conclusion... as we pointed out in the report. But that point was just too subtle to get into these articles." Too subtle to get into the articles? Any reporter or analyst worth a dime could conclude that a sample of less than .002% is inconclusive. Bloomberg quietly altered their reporting after Comscore joined in with a NON-Apple supported study which claims that "iTunes music sales rose 84 percent in the first nine months of 2006 from a year earlier." So what's the real story? Only Apple knows.
The odd part for me is that iTunes average song downloads per day increased 10% in the 200 day period from Feb 23 to Sept 12, 2006 over the previous period (averaged 220 days across Christmas) which included the post Christmas 2005 download frenzy. Sales rates increased 10% for the period these folks are claiming they dropped. Of course, we can't see the trend during that period. As Apple entered this period sixty days after Christmas, were sales still riding the post-Christmas wave only to decline? Or had the post-Christmas wave already crashed with sales now on a steady rise with new iPod owners?
David Caulton, recently joined the debate about iTunes Store's success. David says he has used only Apple's public information to obtain his findings. I find a few faults with his methods.
First, David believes that the success of the iTunes Store should depend on iPod sales. Actually, if anything it should be measured against iPod owners since David says his research shows that 15% of iPods belong to owners of multiple iPods. There is an obvious connection between the iPod and iTunes, yet there is no reason one could not be successful without the other. There is not a single reason that songs have to be purchased from iTunes, it's simply an option among dozens of methods for obtaining music. Furthermore, I know over a dozen people who are/were buying music from iTunes, sans iPod. I also know dozens of people with iPods who have never bought a song from iTunes Store - many buy an iPod just to hold their entire CD collection. The device and the store can exist independently. For example, Napster, Yahoo and the now shuttered Microsoft Music store all operated without a required device. And, Creative, iRiver and many others all sell devices without a requirement for an online music store. iTunes Store and iPod only compliment each other, they are not dependent on each other. Only Apple knows how many songs each individual buys, their patterns and whether they claim to own an iPod or two or more. Still, David has linked the iTunes sales to iPod sales to show a downward trend.
Second, David has pointed out that from Oct 25, 2005 to Jan 10, 2006 Apple sold 3.29M tunes per day, from Jan 10 to Feb 7, 2006 Apple sold 3.7M tunes per day and from Feb 7 to Feb 23, 2006 Apple sold 3.125M tunes. All these periods are affected by the post Christmas iTunes download bonanza. What kind of songs-per-day records does iTunes set on Christmas Day, 5M? 10M? Likely higher. Based on the the trend toward Christmas 2005 (ending at Oct 25), iTunes was selling around 1M songs per day. But then from Oct 25, 2005 to Jan 10, 2006 Apple sold 3.29M songs per day. Most recently, sales have settled to 2.5M per day which includes the after Christmas addition of 14M iPods. We can use a graph to approximate that iTunes was selling less than 1.5M per day up until Christmas. Therefore, within 15 days a 60-day running average of 1.5M songs per day was increased to 3.29M. That requires extremely high numbers on Christmas. iTunes reached such large numbers this Christmas that iTunes was hit or miss for most of the day. That extremely high Christmas Day 2005 number quickly fell to 3.7M two weeks after Christmas and 3.125M four weeks later. With such a large increase in sales directly related to Christmas Day, can we get anything more than a general trend from these numbers?
iTunes song milestones are listed at the end.
Third, I have used the same data to create a graph that very much resembles David's graph of iTunes Songs per iPod. My graph has slight variations and for very good reason. Data is only available in periods as long as 220 days. Obtaining sales curves cannot be done with any degree of accuracy. Furthermore, the dates of iPod sales are released every quarter, which cover a period of ~90 days. Finally, none of the figures have been released on the same dates. So now we have an approximation of 220 days of iTunes sales to an approximation of 90 days of iPod sales. Of those numbers, the reported periods are offset by weeks. I think most people can see the large room for error here. First, iPods could accumulate through the quarter like any of the following graphs.
How do you decide which one to use to calculate with iTunes sales? Then there are the holiday sales which likely trend as so,
I realize you're probably thinking, according to this whole debate's theory, that the iTunes sales should be declining as we approach Christmas day. While this could be true, there's also the fact that from Feb 24, 2006 to Sept 12, 2006 iTunes was selling an average of 2.5 Million songs per day; the highest non-Christmas average daily sales in its history. This was a full 60-days after the post-Christmas song rush.
Next, is the problem with averaging. Because the song sales are averaged over longer periods than the iPod sales, the iPods receive an increase much more often. This means that positive trends of iPod sales are calculated with older and lower averages of songs sales. As with my results, David's sales change direction wildly at the point of public sales announcements. The true graph is likely much more rounded, never reaching the highs and lows that the graph represents. Because of averaging and offset dates, songs per iPod usually wane toward the approaching public song milestone release. They then immediately increase as new higher song sales averages are compared with older iPod averages until the next iPod data is released. There is also the case of the huge peak in iPod sales leading up to Christmas followed by a huge spike in iTunes song sales immediately following Christmas. Again, there are too many unknowns to accurately build the graph David presents.
Next, we'll look at the sales data from a slightly different perspective.
This graph shows us a few things. One, iPod and iTunes sales have trended nicely. Both continue to demonstrate growth, although most recently growth appears be slowing. The most recent data indicates that iTunes music sales seem to be growing slightly faster than iPod sales. This will change during the holiday quarter. Now focus on the red line indicating the total number of iTunes songs sold per iPod. The dots indicate when song sales data was released. Notice the slow down or decrease in songs/iPod each holiday period, highlighted in yellow. This is due to sudden surge in iPod sales during the holiday quarter. Then notice the slow increase of songs to iPods as the next holiday season approaches. Guess what, it probably doesn't really look like that.
This graph doesn't represent the situation any better than David's graph. To create this graph we simply plotted the numbers that Apple provided. The problem still lies in the fact that the numbers are released over too long of periods and not released at the same time. For example, July 17, 2005 to Feb 23, 2006 represents one period of iTunes song sales yet it stretches through three quarters of iPod sales. It also begins and ends inside those quarters. The true graph likely shows a slower song to iPod slope all the way to Thanksgiving day. Followed by a drastic decrease up to Christmas day. Then a very quick rise in songs to iPod sales immediately following Christmas. Apple admits to overwhelming Christmas day iTunes sales.
Because we are only plotting points and not trying to average down to a monthly or worse, daily basis, we can see that ending Sept 30, 2006, Apple had sold 67M iPods and as of Sept 12, 2006 they sold 1.5B songs. This allows us to approximate that there are an average of 22 songs per iPod. There are likely more since we can add 18 more days of songs sales. At an average of 2.5M per day (we know the dangers of averaging this long of a period), and counting David's theory that 15% of iPods are owned by the same person we can come to a final tally of 1.545B songs and 57.5M iPod owners for an average of almost 27 songs per iPod owner. But, since we don't know how many people buy CDs to fill their iPods and how many buy online music, we can't conclude anything about buying trends. Looking back at the graph, since the iTunes milestone for Holiday period 2004 was released immediately before Christmas, it is wildly affected by the holiday quarter iPod sales. Then since, the next quarter includes holiday iTunes sales and post-holiday quarter iPod sales it is also wildly affected. Accounting for these inconvenient offsets, it is likely that Apple has maintained a steady 22 songs per iPod ratio or a 27 songs per owner ratio since late 2004. Still, only Apple knows the way this data really looks...
Finally, let's look at my best approximation of the iTunes sales trend. There is no data to show a decline in sales. I have averaged across the holiday 2005 period. All the sales are still there; only the spike has been removed. I'm not looking for day-to-day or even month-to-month trends. I want to see what the iTunes store is doing. For this we see a clear decline in growth (leveling-off) which mirrors the slowing growth of iPods, but there is not enough data to prove a decline in sales. We will have to wait for future data to determine if there is a true decline in sales. Once a decline is identified, only a major change to the iTunes store would likely change it - perhaps a universal DRm or DRM-free songs. Apple and the music industry will have to discover what exactly is holding people back.
I suspect we will continue to see the same trend for at least another year. That is, a slow but steady rise up to Christmas Day then a sharp rise followed by another slow steady climb to Christmas Day. I don't expect to see a sudden mid-year upswing unless Apple or the RIAA can change the rules. To maintain growth, Apple needs to sell 500 million songs in under 200 days. The clock started on Sept 13, 2006. The cut-off date is March 31, 2007. With holiday sales to be included in those numbers, I expect to see the aforementioned sharp rise. And guess what, that won't answer any questions.
More recent posts indicate that David is more concerned with the music industry's situation. He's concerned that digital sales will never make up for the decrease in physical sales. Did we ever assume it would? CDs force people to buy songs in lumps of 10-20 songs. How many CDs have you bought for one, two or if you're lucky 3 songs? Sure there are many CDs worth owning the whole album. However, most of my collection was purchased for the few songs I wanted. When I got married, my wife was in the same boat. As I ripped her CDs into iTunes she put sticky notes on each CD that read, "Tracks 1,3 and 9" or "Only Track 5." Now that I can buy single-serving songs how often will I buy the CD? I admit I became fans of several bands after hearing their other tracks on CD. I may lose out on experiencing some great bands with single track downloads. I think piracy picked up speed because of its single-serving nature. Oh, and it's free. So with people buying 2-3 songs at a time instead of ~ 12, will digital ever make up for physical losses? No. I used to pirate a lot of music. Two songs weren't worth $14.99 and there was no other legal option. I've replaced everything I pirated and more...
Now I can sleep at night...
Data Used:
04/28/2003
Store Opens
05/15/2003
1 Million/18 days
52631 songs/day
06/23/2003
5 Million/39 days
102564 songs/day
09/8/2003
10 Million/76 days
65789 songs/day
12/15/2003
25 Million/107 days
140186 songs/day
03/15/2004
50 Million/90 days
277777 songs/day
07/12/2004
100 Million/128 days
390625 songs/day
12/16/2004
200 Million/156 days
641025 songs/day
07/17/2005
500 Million/212 days
1415094 songs/day
02/23/2006
1 Billion/220 days
2272727 songs/day
09/12/2006
1.5 Billion/200 days
2500000 songs/day
Comments and Opinions:
0 showing
Post a Comment