Gartner smartphone predictions wrong - predictably

Last September we wrote that we just did not believe Gartner's forecasts for the smartphone market (see Dumb Forecast for Smart Phones):

I had just read Gartners 3 year forecast for the Smartphone market this morning .... and was marvelling at how they could even deign to forecast such a thing with one set of numbers, never mind assume that Android will continue to rise inexorably for the next 3 years just because it has for one year already - haven't they heard of "S" curves?

I was going to write a suitably pedagogical post about the importance of standard deviations around numbers over time, about the inevitability of the S curve, about the need for scenarios, dynamic models and other methods of dealing with uncertainty in this sort of work - but ZD Net has done a good debunking job already (Ed - we were not the only ones who found the predictions hard to believe]

....

An aside - we traditionally (for the last 10 years anyway) take forecasts like these from Planet Mobile analysts and halve them, turns out to remarkably accurate (in an 80/20 sense) for about 5 minutes work

Turns out they have radicaly revised them more in line with our views, ie much more Apple, much less Nokia and Android - Apple Insider.

Gartner's most recent prediction is an acknowledgment that it had no real notion of what was going to happen just 18 months ago. Gartner's original prediction in late 2009 essentially just plotted continued, conservative growth paths out for Apple, RIM and Palm, the three major vendors who own their own operating system. It drew tremendous growth in sales for Nokia's platform, likely due to the company releasing Symbian as an open source project, even though Nokia continued to be the primary (and almost exclusive) user of Symbian.

Android was also given a dramatic projection of unit growth of nearly 11x over three years, as it was just beginning to gain traction. In late 2009, Verizon Wireless was gearing up to replace its former dependance upon RIM's BlackBerry platform with new Android-based partnerships with HTC and Motorola under its "Droid" branded campaign.

However, the biggest and most difficult to fathom prediction by Gartner was Microsoft's reversal of fortune for Windows Mobile. That platform had already clearly lost its initial momentum, with sales remaining flat since the iPhone debuted in 2007. Despite those dim prospects, while Gartner suggested in 2009 that RIM's sales would nearly double and Apple's would almost triple, it stated that Microsoft's would grow more than 4.5x in the same three year period.

Rather than that happening, Microsoft's Windows Mobile platform continue to decline through 2010 despite a heavily promoted deal with LG. And when Microsoft released WP7 as its improved new version of the platform and ended compatibility with previous versions of Windows Mobile, sales tanked completely.

Gartner's 2009 predictions were essentially a bet that openly licensed platforms would make tremendous gains in the market while Apple and RIM would only make incremental progress. While Gartner was correct in guessing that Android would make big gains, it also stated that both Microsoft and Nokia would also grow their platforms at a dramatic rate.

Whereas we have argued till we are blue in the face that (i) Olde style Planet Mobile suppliers (read Nokia) don't have the DNA to build smartphones, and that "Open" (aka Android) is not what it is about, the lessons from the PC market and most others after is that "Software" - or "Apps" as they call them in SmartPhoneSpeak - is what drives intelligent device sales, not device features. And Apple has the winner take all end-to-end supply chain for that.

So why the error? Apple Insider reckons its because of Payola:

If Gartner's historical predictions were more accurate, it would be harder to suggest that the firm was simply concocting its numbers to fit a particular outcome for profit rather than modeling numbers to deliver a useful outlook for the market. It certainly wouldn't be the first time. Microsoft's confidential memos leaked during its monopoly trial state that the company paid hundreds of thousands of dollars to Gartner as it "lobbied" the firm to change its outlook to flatter Windows NT and denigrate competing Network Computers in the late 90s prior to that trial.

More recently, Gartner analysts began predicting in 2005 that Windows Mobile 5 would cause big trouble for Symbian, something that never happened. In 2009, the outlook for Windows Mobile was so bleak that even Gartner refused to say it would surpass the iPhone in three years. The situation for Microsoft has only grown worse over the last year and a half, with the company's Nokia deal seeing little positive comment in the tech media.

Gartner's latest prediction now claims Microsoft's WP7 will grow by more than 1,790 percent over the next five years, significantly faster than Android or the iPhone over the same time period, faster than Android grew last year, and faster even than Gartner incorrectly predicted Windows Mobile would grow back in 2009.

Like Apple Insider, I just cannot see where they are getting their Microsoft assumptions from, it seems hugely optimistic - the "what you have to believes" are fairly extreme in this case. Unlike Apple Insider, I think they are capable of making the errors without payola, because their predictive models do not seem to look at the end-to-end supply chain and how it impacts the game theory of decisions downstream at the device end. In my experience, until a market is fairly mature and just chugging along so you can run it on "last 3 years mean growth", it is essential to understand the "whole system" dynamic.

(I should add that in may other areas I find Gartner quite reasonable, but they have always been very optmistic about the Olde Planet Mobile growth data over the 10 or so years that I've been watching the industry - and they are not the only ones, mobile seems to be continually subject to overoptimism by many analysts - pre smartphone days we used to take a mean of all the major analysts' forecasts and halve the growth - turned out to be surprisingy good 2-3+ years out )