The number of 3G iPhones "sold" in the quarter derived from the IMEI numbers has taken a step increase since 9/26/04 when I posted them here:
Message 24995743
The spreadsheet itself is here:
spreadsheets.google.com
That's the URL link to the spreadsheet I posted on 9/26/04, since for some reason my link allows one to download and save the spreadsheets they have generated as Excel or Adobe pdf files. The link in the article you referenced does not seem to allow downloading.
The current spreadsheet does show 9.2 million 3G iPhones as of the "end date" of 10/5/08, where "end date" = "manufactured and sold" date. These IMEI numbers used in the spreadsheet are taken from actual phones in the hands of end users.
Note that the 9.2 million (actually assigned sequential number 9,190,680 for a unit "sold" on 10/4/08) is the highest sequential IMEI number recorded in the spreadsheet for a unit actually in someone's hand, and some allowance has to be made for IMEI numbers on iPhones still sitting on retail shelves and as yet unsold.
With the "unlikely assumption" that up to 1.5 million of these 9.2 million units are still sitting on retail shelves, the article concludes that at least 7.6 million units were actually "sold" to customers in the quarter. 9.2 minus 1.5 is 7.7 million on my advanced calculator, but whatever. The article then adds in the 2.4 million first gen iPhones sold, and concludes that as of the end of September, Apple had already met the 10 million Jobs projection for the entire year.
Now when I looked at this spreadsheet 10 days ago on 9/26/08, it showed 5.6 million 3G iPhones "sold" as of 8/30/08 (for those interested the full number was 5,604,900, IMEI 1171400604900, Serial 83834), and 6.8 million "sold" as of 9/24/08, the last entry at that time (full number 6,844,000, IMEI 1174300844000, Serial 86837).
From that, I projected 7.2 million 3G iPhones "sold" for the quarter, well above current analyst estimates which I also listed in my post. But I got that number from the reported IMEI numbers, and I made no deduction for "unsold" units, which maybe I should have done - but see the end of this post for a bit more on this
The current spreadsheet shows that same 5,604,900 million sequential number for the 8/30/08 unit (IMEI and Serial as above), but the sequential number assigned to the 9/24/08 unit above (IMEI 1174300844000, Serial 86837), which was assigned sequential unit number 6,844,000, is now 7,844,000.
The reason for this change appears to be that they have injected 12 additional IMEI numbers between the 8/30/08 and 9/24/08 entries I looked at just 10 days ago, which has had the effect of boosting the assigned sequential number of that 9/24/08 unit by 1 million. And they've added 3 more IMEI numbers at the end to get to 9,190,680 as of 10/4/08.
So, how reliable is this estimate of iPhone unit sales derived from the IMEI numbers? It appears to me that it's highly dependent on the sample of the IMEI numbers being a random and representative sample of the total shipped. I think this is a possible flaw in the methodology, since the reporting and addition of just 12 intervening IMEI numbers in 10 days boosted the estimated sequential sales number for the quarter by 1 million.
I would assume that a random and representative sample reporting of the IMEI numbers from the universe of iPhones sold would result in this estimate being accurate, but they are using a sample of only 159 iPhone IMEI numbers out of the total 9.2 million units. It's an interesting and innovative approach, but quite an extrapolation. I'll have to think a bit more about the statistical validity of what they are doing and the possible errors in the estimate.
Also, since sequential IMEI numbers missing from this spreadsheet reflect both "unsold" and "sold but unreported" IMEI numbers, the derived sales number of 9,190,680 is possibly an overstatement of the number actually sold to customers.
However, one question that arises from all this is how Apple "books" revenue from iPhone sales into its financials. Here's what the 10Q and 10K say:
The Company recognizes revenue when persuasive evidence of an arrangement exists, delivery has occurred, the sales price is fixed or determinable, and collection is probable. Product is considered delivered to the customer once it has been shipped and title and risk of loss have been transferred. For most of the Company’s product sales, these criteria are met at the time the product is shipped.
Sounds to me from this that revenue gets booked when an iPhone ships from the factory, and if this is true, the reported sales in the quarterly reports include unsold units still sitting in retail stores, i.e. the channel inventory.
In this case, the unit sales derived from the IMEI numbers may be close to what Apple actually reports, as long as the set of IMEI numbers used is a representative sample of the total shipped and sold through the date in question.
But how good this approach is will not be evident until we see what Apple actually reports in the Q4 financial release.
Until then, since I've now seen how much the sales estimate can change with the incorporation of a limited amount of additional IMEI numbers from sold units, and since I now appreciate that missing IMEI numbers can reflect "unsold" as well as "sold but unreported" units, I'll personally treat these estimates with a bit more circumspection than I did previously.
David |
