Does anybody really know what smartphone market share is?

An article in Wired online this week had a chart showing US smartphone market share (link). The chart gave more detail than I've seen recently from other research companies, so I thought it was worth reproducing the data here:



The source is Nielsen Mobile, formerly Telephia. The interesting thing about them is that they do much of the service quality monitoring for the operators, so they have much more direct access to mobile usage information than folks like IDC and Canalys, the people usually quoted for smartphone share.

Wired was focused on Palm's loss of market share, which is indeed striking (but not exactly news). But take a look at the chart again; there are a couple of other items that I think are more newsworthy.

The first surprise is that Nielsen shows Apple in fourth place in smartphone share. That's wildly different from what Canalys, the source usually reported, has been saying (link). Here's how they compare for Q4 2007:




What in the world is going on here?

I'm not sure, but I have some guesses. Canalys doesn't directly measure market share, it receives self-reported shipment reports from the manufacturers and then adds them up. That means Canalys measures shipments into the channel rather than sales, and it depends on the hardware companies to be honest.

Riiiight.

Nielsen Mobile doesn't explain on its website exactly how it measures share, but apparently it's using a mix of survey results and the usage data it gathers from the operators (link). So its numbers should reflect current usage of phones rather than shipments. If Nielsen is measuring installed base share, rather than share of current sales, that might explain the difference. Although in that case, share should not be changing as fast as Nielsen shows. So I'm still confused.

If anybody can shed more light on the source of the difference, please post a comment. I've also asked Nielsen, and will let you know if I hear anything.

The conflict in the numbers underlines how ridiculously useless the publicly-available third party sales numbers are in the mobile phone market, and how little attention the press is paying to the inconsistencies. Apple's share varies from 8% to 28%, and no one even notices. Hey, we got a pretty chart and it confirms what we wanted to say, so don't ask questions.

If you want more information on the problems with mobile market share tracking, I wrote a detailed post here (link).

I said there were two newsworthy things about the Nielsen numbers. Can you spot the second one?

That's right, since the iPhone was released, RIM has been gaining share. So much for the folks who predicted at the launch of the iPhone that it was going to take the smartphone market away from RIM. Instead, at least in the first round of competition, we see what you'd expect from a segmented market -- RIM appeals to some customers, Apple appeals to a different group, and both companies do well.

I can't wait to see what the numbers will look like in six months, after the iPhone 3G has been out for a while. Although probably Canalys and Nielsen will still disagree wildly on what's happening.

20 comments:

Ben Combee said...

I'd be really curious to see what the chart looks like it it were done as a total number of phones in use rather than market share. I'd expect most of the lines to at least be flat or increasing. Palm keeps selling as many units or more than previous quarters, and while a fair number are replacement, I think we're seeing a pie that's getting quite a bit bigger due to RIM and Apple, but with the relative size of the pieces changing, although few pieces are getting smaller.

Jason Devitt said...

There are lots of problems with this data. Where are Samsung (the Blackjack), Motorola (the Q), and Nokia (the N75 on AT&T, whatever about unlocked phones)?

And should we track this by OEM or by OS? Both, clearly, but as a developer I really only care about OS share. By that measure, the total share of Windows Mobile comes close to RIM / BlackBerry / whatever we call their platform. Microsoft doesn't any credit for that. And the market share for Palm OS is not the same as the market share of Palm devices.

Michael Mace said...

Thanks for the comments, and I agree with both of you.

We'd know a lot more if we had units to go with those share numbers. Unfortunately, as far as I can tell, Nielsen hasn't released any.

I honestly don't know what sales have been in the US for the Blackjack and Q compared to all of HTC's products. But one thing I'm pretty sure of is that each of them would dwarf US sales of Nokia's smartphones.

And yeah, I'd like to see OS share as well.

I'm sure you can get all of that if you pay Nielsen several tens of thousands of dollars a year. If you don't, all you get are these little teaser tidbits they release to the press in order to get free publicity.

Anonymous said...

Hello Michael,
I wonder that they think a phone could be smart if it has a keyboard. iPhone is the first smartphone because of 45 nm CPU.
Michael
GeneTechnics

Michael Mace said...

Yikes! One problem with the web is that it's hard to tell when someone's being serious or sarcastic.

I've been through that "what is a smartphone" argument too many times to count, and it's kind of like quantum mechanics -- the more closely you look at the definition, the fuzzier it gets.

I haven't seen anyone define it by the average half-pitch of a memory cell on the processor before, but if that's what you want to use, more power to you.

Anonymous said...

I'd say more *power efficiency* like in the future Intel's Moorestown platform running IA32 instructions of an average notebook - I really want to use it for a Cell PC.

Michael Molin
GeneTechnics Company

Anonymous said...

Canalys used Apple's 2 million iphone shipment numbers in the christmas quarter.

The problem is that AT&T only activated 900,000 iphones in that same quarter.

Most of those 1.1 million "missing" iphones ended up in China and Russia --- so they DON'T belong to US market share numbers.

So Canalys' 28% market share for the iphone --- can suddenly cut by half if they actually used the AT&T activation figures.

daniel said...

I think the previous post has it right in that Canalys reported shipped iPhones as reported by Apple. Many of these weren't activated in the US.

A related problem alluded to earlier is that much as individuals don't agree on smartphone definitions neither do research houses. What Canalys reports as "smartphones" is not what Telephia/Nelso do. In fact I don't know of any research vendors using similar definitions.

Active subscriber share would definitely be the relevant metric. So far the only attempts to do that which I have seen rely on surveys not monitoring and therefore have their own flaws.

Your larger point is the most interesting. This is an obvious problem that is wilfully ignored. We are asked to accept flawed logic at face value if it comes with a chart.

Michael Mace said...

>>I think the previous post has it right in that Canalys reported shipped iPhones as reported by Apple. Many of these weren't activated in the US.

You're probably both right. Which would mean that iPhone's actual share of the US smartphone market is much lower than the press has reported.

But we can't be sure. I hate these situations where we have to make up explanations to fit the data. I feel like we're letting the analysis companies off the hook -- they ought to be explaining the discrepancies in their studies to us, not the other way around.

But unfortunately Nielsen hasn't responded to the message I sent them, so I can't ask them directly. If anyone else gets a direct answer, please post a note here.

Anonymous said...

I wonder what's the sense to have an iPhone when you have no access to iTunes. Perhaps, to put it on shelf to demonstrate to your guests.

Michael Molin
GeneTechnics Company
--
Surgut, Russia

Anonymous said...

>>>But we can't be sure. I hate these situations where we have to make up explanations to fit the data. I feel like we're letting the analysis companies off the hook -- they ought to be explaining the discrepancies in their studies to us, not the other way around.

No and yes.

No, you are not supposed to make up explanations to fit the data. The analysis companies "CAREFULLY" constructed their wording of their studies. They put a million qualifiers in their studies. It is factually correct that the iphone has a 28% of the US shipment of the smartphone market.

Yes, you are letting the analysis companies off the hook. They put a million qualifiers in their studies and reporters don't even look at these qualifiers.

Remember Ben Affleck's movie "Daredevil" --- it has the best action movie openning in Hollywood history in February. Well, look at the qualifiers --- nobody has ever open an action movie on Valentines day before.

Open a newspaper and go to the movies section --- every movie is number 1 in something. One movie is the number 1 comedy. Another movie is the number 1 romatic comedy. A third movie is the number teen romatic comedy. In the summer time, these don't mean anything --- because all the top 10 movies are big budget action movies.

steve hawker said...

Can't resist this thread :)

As interesting as market share figures are (inaccuracies and hyperbole aside), I find purchase intentions to be much more indicative of the state of the market because it shows the trend of things to come, not the trend of things that have passed.

To whit, ChangeWave have just release their quarterly research. Here's a link to some data:

http://www.cellular-news.com/story/33266.php
(click on the small graph)

Note the wonderful hyperbole - Cellular News are certainly hypenotized!! But there's some really interesting stuff here - just as Mike says, the journalists have hidden the lead - TWICE!

Palm's share of purchase intention has plummeted from 22% to a mere 6%. Shouldn't the lead be "Palm ignored by business users"? Yes, I know the figures for Centro are record breaking (by Palm standards) but all this means is that Palm has lost the high value business market and has moved to the ultra-competitive consumer market (I eschew the term prosumer). Imagine Palm's position if Jobs really does bring out a Nano iPhone for $99...

The second lead is that way back in Feb 07, 17% indicated they would purchase some other smartphone, holding steady in May 07 at 16% - but since the launch of the iPhone no-one is interested in anything but RIM, Palm and Apple (so there is some good news for Palm). What's that you say? In the last five quarters the purchase intention for WinMo devices is zero. Zero. (OK, some months the numbers add up to more than 100% meaning perhaps that some folks were returning their WinMo devices?).

This is truly staggering. Despite all the effort Microsoft has been to, WinMo is looking very dead. Indeed, as Mike says, I definitely look forward to market share numbers in 6 months, and more, as WinMo users increasingly throw their impossible-to-use devices in the garbage and buy either the sexy iPhone or the workhorse RIM. Really, the only question is: can Palm resurrect itself with another industry changing device (for the 3rd time...)

Either way, its fun watching!!

Anonymous said...

Steven?

HTC sells windows mobile devices and the chart on top said that it's in second place.

Surveys of "intended purchases" means nothing. They actually have to buy them first. It is especially true in the smartphone market where the consumers don't realize how much data plans cost per month.

steve hawker said...

Anon?

Well, we can go round and round on this...

I'm turned towards intentions because that reflects what is in the mind of the purchaser and it does reflect the share of mind that a particular device or vendor is getting at the time. Hardly "nothing" when vendors spend so much money trying to get that share of mind.

HTC may well be second in the so-called market share but note how their "share" has been falling over the last few quarters - reflecting how their sales have been hit by the realised lack of purchase intent! Same for Palm.

Obviously purchase intention is just that - an intent. But device vendors shouldn't ignore this data. Devices take many months to develop and if HTC (and Palm) are just sitting there congratulating themselves on how strong their market share is, then by the time they realise their share is no longer as strong, they're 2 years behind the competition.

And yes, some folks are turned off by the cost of data plans, but its somewhat of a red herring. Despite the cost, some folks still buy smartphones. I've heard some stats bandied around that suggest that as many as 30% of smartphone owners don't even have a data plan - makes you wonder, doesn't it.

Anonymous said...

Stephen,

You have to look at how ChangeWave asked the questions in the first place.

The survey was done --- before AT&T announced their price plans.

http://www.changewave.com/freecontent/viewalliance.html?source=/freecontent/2008/08/alliance-research-in-motions-3G-counteroffensive-08-11-08.html#top

http://www.apple.com/pr/library/2007/06/26plans.html

Of course, their numbers shot up like a lot --- all they heard from Jobs was "half the price".

As with HTC --- we have been talking about how inaccurate the market share are for the Apple iphone in these surveys (because 1/2 of the US iphones were bought by people to sell overseas). It means that the market share drop for HTC and Palm is less than you think.

steve hawker said...

Anonymous,

ChangeWave's survey is published quarterly so it encompasses all the plan and device price changes over the last 2 years. Plan prices change often - pointing to a single change is a red herring.

Of course price is going to affect people's purchase intentions - I'm not sure what your point is here. In a free market Palm, HTC, Mot etc are all free to set their pricing to be competitive and if the carriers aren't pushing certain devices, then that's another artifact of a free market. If people can hear Jobs say "half price" what's to stop people hearing Palm, Mot, HTC etc saying something similar?

The fact is that RIM and Apple are steadily making the smartphone market a two horse race (in the US for sure, Europe still has Nokia being relevant). And its not a surprise...

Regardless of whether Palm or HTC's market share (of shipped product) is bigger or smaller, and regardless of whether people are buying iPhones to ship elsewhere, folks are stating loud and clear that they are only interested in RIM and Apple.

Anonymous said...

Other manufacturers can never advertise like Steve Jobs. All you hear is half the price for the iphone.

Also intention surveys don't matter. Remember all those surveys done before the first generation of iphone were sold in Europe --- how many millions of people were signalling their intention to buy in UK, Germany and France.

Then the reality set in --- nobody bought them in Europe.

How is it a 2 horse race? Apple's market share drop by 1/2 by Q1 and then it drop further by Q2. All you see is inflated numbers of Q4 in which 1/2 of the iphone went missing.

Anonymous said...

It is also interesting to note that HTC "market share" only includes HTC branded devices in the Gartner reports.

If HTC designs and manufacturers smartphones for mobile carriers, they are listed in the mobile carriers' market share --- not HTC market share.

http://www.gartner.com/it/page.jsp?id=754112

Therefore HTC's real market share is much higher than the market survey suggest.

james wilkins said...

Apple Inc. managed to garner approximately 28 percent of the red hot US smartphone market during the fourth calendar quarter of 2007, up more than 43 percent from the three-month period ending September.According to a report issued Tuesday by market analysis firm Canalys, the iPhone's 28 percent share placed it second in the US market behind only RIM's with 41 percent share, and well ahead of Palm, whose 9 percent share placed it a distant third.
---------------
james wilkins

Link Building

Michael (Amy's pet Kiwi) said...

Are you sure that it's not what the first respondent spoke of... Telephia tracks users, that's installed base. Canalys and most others track sales, which accumulates over time into installed base.

Michael