Hello my readers. You recently were able to see how the contest between Tomi Ahonen and myself ended up. Today I’d like to go a bit deeper into Tomi’s forecast about year 2015.
You probably already know Tomi Ahonen calls himself the most accurate forecaster so I want to see more details on what his forecast for year 2015 actually was.

As Tomi titles himself as mobile analyst he shouldn’t have just pulled the numbers from his hat but actually have market share percentage model available. Unfortunately he did not give us the percentage list but we do have some data points to reverse-engineer the Tomi forecast. Let’s however start with this gem:

Pyramid was widely ridiculed for that worst forecast of the industry and for the record in 2013 Android sold 767 million units of smartphones and Windows Phone managed 33 million. That error is as far as I know the biggest error in any forecast in mobile ever made, at 2,200 percent !!!

–Tomi Ahonen, April 08, 2015 [1]

Tomi’s forecast? off by 7,750 percent – OVER THREE TIMES “the biggest error in any forecast in mobile ever made”. [2]

How do we get there? It’s easy and here goes:

First of all we do know that top 3 mobile operating systems were supposed to be Android, Tizen and iOS (in that order) and they will hold “over 90% of the market”:
tomi_tizen_2013_several [3]
In addition to that Tomi also gave top 5 order [4], which was Android, Tizen, iOS, BlackBerry, Windows (in that order):
He did say it then that Tizen and iOS will be close to each other and he had said that before too [5]:
tizeniosclose

Now this start leaves us with few questions, like:

  • How big are the sizes of Android, Tizen and iOS?
  • We do get maximum market share for Tizen at 30%. Simply because Android needs to be larger, iOS needs to be almost the same size and combined market share needs to be 90% (or a bit over). Basically that would be 31%/30%/29% for Android/Tizen/iOS.
  • We also know that the minimum market share for Tizen is at 6% since the last OS’es take 10% (and if we have maximum of 5%/5% BB/WP we need to have Tizen at least at 6% to be “larger”).
  • But how about the actual ratio between Blackberry and Windows? It defines the minimum Tizen size so is it 5%/5%? Or perhaps 9%/1%?
  • And most of all: should we leave anything for “other OS’es”?

There are too many variables so let’s see what we have to help us. First of all we know Tizen is supposed to be huge. It’ll be bigger than iOS but “probably” never catches Android:
tizen_will_outgrow_iOS_probably_not_Android

And we also know that Tizen will eat about 10 points of percentage of market share from Android and Android would then be at 60%:

Note, that most of Tizen’s gains would come directly from Android. Still, if you remove 10% from Android’s 70% now, it still leaves Google as the Gorilla of mobile, with 60%

–Tomi Ahonen January 31, 2013 [6]
Androidpeak
[7]
That statement was made at the same time as Tomi’s previous forecast, so we should use highest Android market share for year 2012, which is 70.7% from Q3 (same as Tomi mentioned), assume it was “the peak” and minus the 10 points. (This increased Tizen minimum market share to 10%, BTW.)
At least Tomi was fine with the idea that Android market share will be down. Let’s for now leave Android there at 70.7% minus 10 so we’ll have it a bit above his 60% prediction.
Next I’ll put Tizen just 0.1% ahead of iOS as those were supposed to be close (and since in reality Tizen did not outgrow iOS, this tries to minimize the error made by Tomi).

More data points: Blackberry was at 3% and Tomi said they will remain niche [8][9]:
blackberry3percentniche
So I’d count that 3% for BlackBerry. We can’t really have lower number for poor BlackBerry or “others” corner will be problematic (more about that next).
Windows Phone/Mobile Tomi said will never pass 2%. [10]
https://twitter.com/tomiahonen/status/262922925978566656
In addition to that Tomi listed SailfishOS, FirefoxOS and Ubuntu as something coming in the future. And the fact is that all of them must be smaller than Windows or otherwise Tomi’s top 5 ordering is lost. And none of them should be noticeably bigger than the others but Sailfish OS is the one with 1% market share. [6]

Let’s gather the market share requirements:

  • Order is Android, Tizen, iOS, BlackBerry, Windows.
  • Android has ~60.7%.
  • Difference between Tizen & iOS is as small as possible, Tizen leads.
  • BlackBerry has ~3%, not more.
  • Windows can’t have more than 2%.
  • Sailfish OS, Firefox OS and Ubuntu are smaller than Windows.
  • Sailfish OS has 1% market share.

Now we could go for 60.7% Android and split the rest of 90% equally between Tizen and iOS. That would be pretty bad forecast for iOS but even worse for Android if we tried to increase iOS market share to meet the real iOS result from 2015.
For the rest we should take maximum sizes for BlackBerry and Windows (3% and 2%) but then we have 5% to spread between the other OSes, of which none can be over 1%. And we cannot increase Windows or BlackBerry OS market sahres either so let’s try this one:
Tomi said Android, Tizen and iOS will be OVER 90%. Let’s toss that sum to 92%. We know Android is supposed to have peak at 70% and then lose 10 points to Tizen so let’s not increase Android market share (as we shouldn’t or iOS is off again).This means Tizen and iOS gain the last 2 points. We cannot make them differ much (and we shouldn’t if we want to keep this as close to reality as possible).
Let’s add one more point: Still in 2014 Tomi said that in his past vision Samsung would be shipping half of their devices running Tizen by 2014 already, plus there would be other vendors included. [11] Samsung market share for 2014 was 24% (down from 31% in 2013) which means Tizen needs to be around 15% at least.

I think we are ready to have the full set and it looks like this:

  • 60.7% Android
  • 15.7% Tizen
  • 15.6% iOS
  • 3% BlackBerry OS
  • 2% Windows
  • 3% others (Incl. Salfish OS, Firefox OS and Ubuntu)
  • 0% Symbian

Now the actual full year 2015 results: [2]

  • 81.3% Android
  • 16.1% iOS
  • 2.0% Windows
  • 0.3% BlackBerry OS
  • 0.2% Tizen
  • 0.1% Others (Incl. Sailfish OS, Firefox OS and Ubuntu)
  • 0% Symbian

Tomi is somewhere in the other galaxy with Tizen, clearly had no clue on how Android would be de-facto OS and was not on track with BlackBerry either. I think this explains why Tomi only talks about Windows Phone when he tells he’s “the most accurate forecaster”. 😉

Could we compare it to something? I know! Let’s compare it to the worst forecast we can find! That is of course the one Tomi calls the worst of them all and comes from IDC:

  • 52.9% Android
  • 19.2% Windows
  • 19.0% iOS
  • 5.9% BlackBerry
  • 3.0% Others (incl. Tizen)
  • 0.0% Symbian

Tomi scores closer on Android, iOS, Windows and BlackBerry but fails more with every other OS, especially with Tizen. But let’s score this properly (back to the original punchline in my post):

For Windows Phone forecast Tomi said Pyramid forecast of Windows Phone is off by 2,200% because their forecast was 23 times as big as the reality turned out to be. [1][12] (Idiotic way to compare the forecasts but that’s the style Tomi wants to use.)
Fine.
With the same math Tomi’s Tizen forecast is off by whopping 7,750% – over three times worse than Pyramid ever was.

If we want to be REALLY merciful we can ignore everything Tomi said about peak Android, Samsung shipping half of its phones with Tizen, Tizen challenging Android and so forth and push Tomi’s forecast all the way down to the 10% absolute minimum market share Tomi clearly stated Tizen would exceed (inconveniently putting iOS at 9.9%). But even then Tomi is off by 4,900% – over TWICE as bad as Pyramid.

That error is as far as I know the biggest error in any forecast in mobile ever made

REFERENCES:

[1] http://communities-dominate.blogs.com/brands/2015/04/reviewing-my-last-4-year-forecast-blog-3-years-in-from-2012.html
[2] http://communities-dominate.blogs.com/brands/2016/03/smartphone-wars-full-year-2015-final-top-10-by-brands-top-5-by-os-installed-base-dec-2015-and-q4-dat.html
[3] https://twitter.com/tomiahonen/status/249290909684998144
[4] https://dominiescommunicate.wordpress.com/2016/03/14/the-competition-for-most-accurate-forecaster-is-settled/
[5] https://twitter.com/tomiahonen/status/276412864657489921
[6] http://communities-dominate.blogs.com/brands/2013/01/early-thoughts-on-new-operating-systems-ubuntu-sailfish-jollameego-tizen-samsungs-update-to-meego-fi.html
[7] https://twitter.com/tomiahonen/status/306278660351074304
[8] https://twitter.com/tomiahonen/status/296658032002220032
[9] https://twitter.com/tomiahonen/status/297264889263697920
[10] https://twitter.com/tomiahonen/status/262922925978566656
[11] http://communities-dominate.blogs.com/brands/2014/10/iphone-393-m-units-in-q3-sorry-i-missed-this-news-a-week-ago.html
[12] http://communities-dominate.blogs.com/brands/2013/01/final-score-on-most-important-forecast-of-2011-once-again-tomi-ahonen-most-accurate-forecaster-in-mo.html

Advertisements