We have seen over 400 news stories cover the finding by Canalys claiming that Android has passed Symbian in Q4 for the best-selling smartphone OS. Canalys counted Google (Android) at 33.3 million smartphones sold in Q4 vs 31.0 for 'Nokia' ie Symbian. This is not true. Canalys has a major error in their math. The story as now reported is false.
I have no problem with the Symbian number, that is well in line what other sources say. But the Google Android number is premature. Yes, Android grew strongly from Q3 to Q4, but the numbers do not support Canalys's view that Android handset makers shipped a total of 33 million handsets in Q4. Google's own announcements do not support that number.
What do I mean. On August 4, Eric Schmidt the CEO announced Google was activating 200,000 Android handsets per day (ie a monthly level of 6 million and quarterly level of 18 million Android devices). That had been 100,000 per day as recently as May 20 according to Google - a very strong growth rate of 26% per month.
Then the next milestone reported by Google didn't happen until December 9, when Andy Rubin of Google reported they had reached the 300,000 activations per day (9 million Android phones per month, or 27 million per quarter).
For the Canalys number to be true, Google would have had to average 370,000 Android activations per day - as an AVERAGE - for the quarter - ie 370,000 level should have happened near November 15 - more than 3 weeks before Google announced the 300,000 level. Or even more bizarrely, if Google was activating 300,000 on December 9, they would have had to sell at the level of over 600,000 per day for the last 3 weeks of December.
We have heard at every milestone from 50,000 activations to 100,000 activations to 200,000 activations to 300,000 activations. The speed of growth has been slowing down towards the end of 2010, not increasing. We will definitely hear when Google passes 400,000 activations and by January 31, 2011, that had not yet happened. My math says it should happen in March.
But to show exactly what is consistent with Google announcements, here are monthly activations, assuming a linear growth rate, at 26% monthly growth from May to August, and 11% growth from August to December:
May . . . . . . . 100,000 per day
June . . . . . . . 126,000
July . . . . . . . 159,000
August . . . . . 200,000
September . . 221,000
October . . . . 245,000
November . . . 271,000
December . . 300,000
January . . . . 333,000
These give a quarterly level for Q3 of 18.7 Million and for Q4 of 25.8 Million. I can easily accept some error of the magnitude of 10% or so, so the Q4 number could be in the 28 Million range but no way could it be 33 Million. That is an error of 28% above what Google reports. That is no way reasonable.
So lets do a double-check. If we add all Android provider total handset sales, and remove Microsoft Phone 7 and Microsoft Windows Mobile sales; and remove Samsung's bada sales - we get 29 Million sales. This is the ceiling - some of those providers also made some Symbian handsets still, and some Linux Mobile etc phones. No, 33 Million is simply wrong. The reality is somewhere between 26 Million and 29 Million.
I do fully acknowledge that Android is growing strongly, and will shortly pass Symbian - because of Nokia's recent catastrophic decline in market share - but that point has not happened yet. I hate it when an analyst makes such a blatant error - where all facts do not suppor the finding - and they make a headline about this, and get big media coverage for something which is not true.
I hope that Gartner and IDC, when they report on smartphone sales shortly, will provide us with the correct numbers, because Canalys definitely jumped the gun. Shame on you Canalys - you caused a false story. That is not professional of you.
UPDATE 9 February - Now we have both IDC and Gartner reporting that Nokia's Symbian was still bigger than Android in Q4. Gartner went out of its way in its press release to express this specific point. Who told you first, eh? I think I know my numbers haha..