I was immediately openly critical on Twitter about the first version of GigaOm's analysis of Apple App Store revenues. The first posting was outrageously too high and had clear blatant math errors. GigaOm have apologized on their blog - there are numerous comments on the site who all point out the errors. We don't need to cover those, GigaOm has done the right thing and corrected their math errors.
Now, I am not here to bash App stores again or Apple. I am here to correct bad analytics and statistics as the unofficial stats police for the mobile industry. And even after the corretions, its still bad math.
So currently they say that after all the math is done, the Apple iPhone App Store generated 250 million dollars in revenues just in the month of December. They then have the correction to the math and there is a clarifying explanation on the math used which they say is:
For clarification purposes, here is the math:
According to Flurry, Average listed price of a paid app: $3.63
74% of apps listed in the app store are paid. Average listed price of an app (including free): 3.63 x .74 = 2.70 (with rounding)
Only 1/4 of downloaded apps are paid. Average price paid for an app (including free): 3.63 X .25 = .91
These numbers seem more reasonable. And obviously we don't have deeper insights on how Flurry got to its average listed price of paid app. But I find the number bizarrely high. If half of the Top 50 apps sold have a weighted average price of only $2.59 (counting the exact average of the 50 bestselling app prices listed on the GigaOm graphic) - why would total sales of all iPhone apps be more than that? The Flurry average number of $3.63 is 40% above just the average of the 50 best-selling apps. If this is to reflect all paid apps that were sold they report were 70 million (25% of 280 million during December), to me this is remarkably counter-intuitive. If Flurry (or GigaOm) gives such a bizarre number, of all other apps that are not top 50 most popular apps, the users are willing to pay 40% more - no this does not smell right. Can't be.
I would argue that logic dictates that if you are not a bestselling app, you cannot charge the highest prices. Those not in the top of the list have to ask for less than the bestselling apps? Makes sense. So the average of the top 50 bestselling apps assumed half were priced 99 cents. I am pretty sure that of all downloaded apps on App Store - by download volume - if that is the top 50, then for the rest, there will be more than half which cost only 99 cents. If we assume that is 60% of all downloaded paid apps, the average drops to $2.44 and if we assume 70% of all paid apps downloaded cost 99 cents, then the average drops to $2.32. (other things being equal, obviously)
To make the Flurry math work out - we would have to go the other way Of all downloaded apps, only 17% would cost 99 cents, to make their weird number work out ! Only one in six apps sold on the App Store would cost 99 cents to get the math to work out at an average of about $3.63. This is very very unlikely to be true. I am afraid, that Flurry has only counted the 'price list average' and used it as their estimate of what downloads were paid. The error is like to count the costs of all cars as the average price. So the Toyota price gets the same 'weight' in the average as the Bugatti price. And the average will be far higher than what is the real paid average of a car...
As I said, I do not know. but certainly, we have to assume that the average of the 'top 50 bestselling apps' will be reasonably close to the real average, and furthermore, that by anecdotal evidence all who are not gaining hit apps will drop their prices either to the lowest acceptable - 99 cents - or to free. So the reality will be less than this.
If we adjust the GigaOm numbers to the average of the top 50 bestselling apps of $2.59, and then do the math as explained by GigaOm in the above we get ($2.59 x 0.74 = $1.92) is average listed price of app including free apps. And we get ($2.59 x 0.25 = $0.65) is average price paid for an app including free.
Now, if we assume the 280 million apps number is correct for December and we assume one quarter of those were paid apps ie 70 million, and the paid apps had an average cost of 2.59, the total revenues earned by the Apple App Store in December were not 250 million dollars but 181 million dollars.
The more likely reality is that there are more 99 cent apps of all downloads than in the top 50 hits, so if we take the 60% of all downloaded apps cost 99 cents level, the annual sales revenue figure is $171 million and if we assume 70% of all downloaded paid apps cost 99 cents, the total sales number falls to $162 million dollars. These figures to me are within reason. And obviously Apple gets to keep one third of that number.
In sum, a reasonable analysis of those numbers reported is that the total earned by all Apps sold at the Apple iPhone App Store in December was between 162 milllion dollars and 181 million dollars, not 250 million. Lets call it $171 million for December. And now please do not make the mistake of multiplying $171 by 12 to get 'annual revenues' for the App Store - as that number reflects enormous continuous growth, and a strong seasonal effect (Christmas sales of both the iPhone devices and the Apps themselves, including seasonal Christmas themed apps not sold on other months)
And also if you feel like calculating what the average user pays etc, also remember the reported numbers on the GigaOm site have total shipped iPhones and iPod Touch devices (58M cumulative active users). That number also is way too high. AT&T reported that 43% of all iPhones sold have been activated on the AT&T network - but also AT&T reported that they had one third of active iPhones (11M out of all 33M shipped) in use. So one quarter of all iPhones that AT&T has once activated are no longer used (on their network). Some may have been early iPhones that were smuggled out of the country haha, but the vast majority of those are old iPhone 2G models that have been since replaced by iPhone 3G or iPhone 3GS models. So only about 75% of the installed base of iPhones are in use. And if Flurry used Admob stats to model the 'proportion' of iPod Touch's vs active iPhones, then the real number is not (34M + 24M = 58M) but rather is closer to (25M + 18M = 43M) actively used devices that are either iPhones or iPod Touches. Not 58 Million but 43 Million.
Lets be careful with those stats out there... Now, while the real numbers - based on the data reported on the GigaOm site - are nearer $171 Million dollars of income this December for 43 Million active users, those are very impressive numbers and showing dramatic growth. Why not report the truth and not the silly $250 million/58 million numbers like GigaOm did.
And about this $171M dollar number. It still doesn't 'smell right' to me. I did some quick modelling and found that the total still comes out too high for cumulative sales. I am certain Apple will annnounce the moment their total cumulative sales pass a Billion dollars. And thus my gut says the real number for December was even less than this number. So please don't go around saying 'Tomi Ahonen said App Store generated $171M in December' but rather, that Tomi Ahonen analyzed the GigaOm numbers and came to the conclusion it could not be anywhere near the $250M, and that he thinks after all is said and done, the real number is way less than $171M. Ok?