Update on 23 September 2011: A new report from boingo says that iOS devices dominate use at Airports, as measured by connections to their Wifi.
Checking email will be a major use case at airports, so I think that boingo is seeing the same issue as CampaignMonitor did - iOS downloads images by default, unlike other common email clients, and so its real use is over-reported.
Here’s the original post…
“Users abandoning desktop email clients for mobile, study reports” - TheNextWeb
That report, and others, are based on a survey from CampaignMonitor:
“every time one of your subscribers opens an email you’ve sent, we record which email client they were using … After crunching the numbers from a little over three billion email recipients, here’s how it looked”
… Mobile email clients like the iPhone, iPad and Android have grown from 4% of the market to almost 20% in just two years.
The data is right, but the conclusions about useage are wrong. Here’s why.
Emails opens are not email useage. The issue here is that, when people use a mobile client for their email, this reports a lot more opens than using a desktop or Web client for the same email. Analysts then mistakenly interpret the extra opens as extra useage.
One example (Android).
When I get to the office, I check my emails that have arrived overnight, using Web Gmail from my desktop. I scan through the inbox, reading the subjects, and mark most emails as read without ever opening or reading them. This deals with about half of my daily emails, but I don’t open many emails, and so this generates very few “open” reports.
For the rest of the day, my Android phone alerts me to each new email and I click the alert to mark the email as read. This opens every single email, which must generate a lot more “open” reports. (Note: in both cases, opens are only reported for addresses that I have allowed to display images. So most opens are not reported. The point is that more are reported in the mobile case) But I don’t read any more of them than when using my desktop, so there is no more useage.
And another example (iPhone):
On iPhone, images are read by default when you open an email. This includes the “web bugs” used to record opens. But, on the main Desktop and Web Mail clients, images are not read by default. This means a greater proportion of opens are reported on iPhone.
So, when reports of a survey use opens to estimate market share, as here, this exaggerates the importance of mobile email.
Fortunately, it is simple for future studies to avoid this problem:
Never calculate email market share based on opens. Please ignore any reports that do this.
Use click-through and outcome data instead. The numbers are much lower, but they are more reliable.
Update on 24 June 2011: Let’s try and correct the data.
The market shares from the CampaignMonitor survey look like this:
The iPhone number is 8.73 times the Android number. But in terms of ownership, these platforms are closer: USA Figures (iPhone and Android approximately equal), European Figures (iOS has 2.16x the users of Android).
Let’s do some maths.
I need the European ownership figures for iPhone, not iOS, so I’ll use the ars technica figure that 58% of iOS devices are iPhones. Calculating based on this, in Europe, iPhone has about 2.16 x 58% = 1.25x the users of Android.
I’ll assume the USA is about 1/3 of the market and that the rest of the world buys the same things as Europe. This means that Globally, iPhone has about:
1x0.333 + 1.25x0.667 = 1.167x the users of Android.
The pie chart from Commshare shows much greater numbers for iPhone than this figure suggests. Let’s assume this is due to over-reporting of opens. I have reduced the numbers in proportion to the numbers of devices owned, assuming iPhone and Android users are equally likely to use email. Here is the chart for estimated email opens (not just measured opens):
That looks much more believable than the raw data.
The maths reduced the total number of mobile clicks by 4.81x
Update on 23 September 2011: I want a round number. My maths assumes that iPhone and Android email useage are exactly equal. I think they are close, but other reports say that iPhones are used more, and I have never seen a report that says Android is used more. So I am going to round this number down. So my estimate becomes that mobile useage is over-estimated by 4x in the above survey.
Correcting the quote that I took from the survey:
Mobile email clients like the iPhone, iPad and Android have grown from
4%1% of the market to almost 20%5% in just two years.
Update on 23 September 2011:
- Some other recent email data suggests 9.3% - see Mobile email usage statistics - EmailMonday. But some of this data is suspect, for example see “”Why does Litmus report more opens than my mailing software?”, and the rest is calculated on a different basis (see a guide to mobile statistics) - e.g. counting people who open emails on mobile at least once per month, which I think it too low a level to matter. So I think 5% is still about right.
- The market share of mobile browsers is about 5% - Netmarketshare, so perhaps email has a similar percentage. I don’t see why email would be used less than Web on mobile.
- Smart phone useage is growing very fast, and the iPad is proving a very popular device for emailing. So it can’t be long before mobile email reaches 6%.
OpenOffice Calc Spreadsheet available on request - Pete