Skip to main content

Explaining the iOS and Android mobile browser usage disparity

By Jason Grigsby

Published on August 1st, 2012

Topics

John Gruber mused today on why there is a disparity between the marketshare of Android and its share of web traffic. He theorized that “an awful lot of Android smartphones don’t really get used as smartphones”.

That may be part of the story, but I think there is better explanation, and thanks to Akamai, we have data to support it.

In June, Akamai released Akamai IO which provides us with a view of the web we’ve never had access to before. It aggregates web browsing data from Akamai’s clients.

For those unfamiliar with Akamai, they are a content delivery network that is used by many major web properties. The means that their data is likely to be a more accurate reflection of web traffic than we’ve had from other sources in the past. They see more data and the sites their clients have are the popular, mainstream sites that everyone visits.

Enough background, let’s take a look at some data. First, this is what Akamai shows for mobile web browser market share over the last 30 days:

Akamai graph showing Mobile Safari dominating Android over the last thirty days.

No real surprises here. Mobile Safari leads Android in web browsing usage by a significant margin as others have reported in the past. What gets interesting is when you look at traffic by type of connection. First up, non-cellular networks:

Akamai graph showing Mobile Safari with an even wider margin on non-cellular networks

Again, no big surprises here. If anything, Mobile Safari’s lead over Android has increased. What about if we look at traffic on cellular networks?

Akamai graph showing parity between Mobile Safari and Android on cellular networks

Now that’s interesting isn’t it? iOS and Android are neck and neck when you look at web usage on cellular networks. So the question becomes not why do Android users not browse the web as much as iOS users do, but why don’t they browse the web on Wi-Fi connections?

It is important to note that while Akamai’s data is better than what we’ve had before, it isn’t perfect. On the question of whether or not the data has a bias, Akamai says:

Currently, yes – most of the Web sites sampled for the Akamai IO Beta are focused on a U.S. audience. Therefore, while the data in the reports includes many visits from across the globe, it is biased in favor of U.S. clients. As we grow the sample size, we’ll no longer have this bias, and in turn be able to introduce geo-specific views of the data.

To my mind, the U.S. bias makes the difference in network usage even more striking. Many cities in America have abundant Wi-Fi locations. Why don’t Android owners use those hotspots?

I have a couple of theories on why Android users don’t browse on Wi-Fi as much as iOS users:

  1. The UI for joining a Wi-Fi network on Android is easy to miss. iOS frequently prompts you asking if you would like to join a Wi-Fi network. The prompt to join the network takes up the screen and interrupts what you’re doing. In fact, it can be a tad annoying.
    Android, on the other hand, puts the notification as a subtle indicator in the notification bar. This is easy to miss if you’re focused on other things or if you’re a new Android user—which by the way, many Android users are because the majority of first time smartphone buyers choose Android.
  2. People at lower income levels are less likely to have access to Wi-Fi networks on a regular basis. I believe a lot of differences between iOS and Android can be attributed to income level differences between the users of the devices. Last year, Comscore published data showing that 81% of iPhone owners in the United States were above the median income level.
    I’d love to see more recent data and be able to compare it to Android, but every time I’ve seen demographic data, even if it doesn’t explicitly address income levels, it seems to hint at the fact that Android users have lower household incomes than iOS users.
    People at lower income levels are less likely to also own a computer and therefore less likely to have broadband at home. They are also less likely to work in an office setting with abundant broadband and Wi-Fi.

As an aside, I believe income level disparity also helps explain the differences in app store success. There are many factors, but demographics need to be included when people compare the platforms.

It probably goes without saying, but if you’re on a Wi-Fi network, you probably have a faster connection and you’re usually not worried about the amount of data you use. Having a faster connection and no worries about your bill increases the likelihood you will browse the web more. The friction for browsing goes down tremendously when you’re on Wi-Fi.

Even with Akamai’s data, we’re still just theorizing on what is going on. But I do believe that the data provides the first clue as to what might be happening. And if Google is smart, they’ll dig into this disparity further to see if there is something that can be done to increase the likelihood that Android users will connect to Wi-Fi networks. As outsiders, we can watch these metrics to see how well Google is adapting.

Finally, I’m pleased to have an opportunity to talk about Akamai IO’s data. It’s a great service that they’ve provided. And I know they have plans to make it even better. I suspect it took a lot of work to reassure people that it was ok to publish this information.

So thank you to those inside Akamai who championed it. And for the rest of us, let’s put this data to good use.

Comments

Jason Grigsby (Article Author ) said:

One thing I forgot to mention. Guy Podjarny told me that mobile browsers include tablets. So part of the disparity could the success of the iPad versus Android tablets. The assumption would be that iPads are more likely to both be on Wi-Fi and be used for more browsing.

Jason said:

The iPad’s the real story here. Not only does the iPad dominate the tablet market, but people view more pages on tablets than on smartphones. At my last job, users coming via tablet viewed about 5 times more pages than they did on smartphone (12 pages Vs. 2.5). I’d love to see the numbers once/if the Nexus 7 starts taking off.

Joe said:

If android and iOS cellular usages are about the same while android phones are reported to have 2x ios market shares, it shows android users are 50% less likely to surf the web using cellular network than iOS users. What gives?

KK said:

@Joe: that makes perfect sense and raises my point at the income related theory above. If Android phones are favored by lower income groups, then it would make sense that they would not spend more on data plans and thus after reaching the data limit, they would not access the web.

Replies to KK

chris replied:

I do not think it is “income level”.
A friend is teacher and I gave her my iphone. She never really understood how to do simple tasks like setting alarm(she let the children at her school do it for her) and she was not really interested in learning it. She simply used it for calling people (not even messaging). When her iPhone was stolen she got an Android. I don’t know if she still has her pupils help her with it, but I think, people who are not interested in apps, eBooks, movies, fotos and surfing the web make their choice based on price – and Android phones are cheaper than Ios devices, at least here in Europe.

Maybe people with lower income are also those not so interested in technology – but they want to own a smartphone, even if it is only used for basic phone services. If blackberry would sell theirs cheap they would certainly be very successful.

Kenneth Beger said:

Akamai’s data is interesting but skewed as the sites using Akamai’s distribution network are typically very savvy large companies. Not necessarily representative of general use.

Replies to Kenneth Beger

Jason Grigsby (Article Author ) replied:

What you see as a bias, I see as a strength. Despite how much I like the folks at StatCounter, I’ve not encountered many sites that use their stats page. Same for Net Market Share data which collects its information from clients using HitsLink Analytics and SharePost clients.

I’m pleased that both services provide aggregate data, but what I would really like to know is what the percent of people hitting CNN, Apple, Fox Sports, etc. are using because the volume is much greater and the cross section of users is much greater. Sure there may be a few people who only visit small sites and never visit anyone using Akamai, but they are far and few between.

Replies to Jason Grigsby
Steven Hoober replied:

I wish I worked places that let me talk about the research and stats specifically. And boy do I wish ESPN and CNN and so on would give their stats.

But I’ll tell you from the last few I have seen (and some are quite general products), I do not observe splits this strong. At all. Most places, for most products at the organization, lead with Android. Sometimes, by 2:1 in favor of Android browsing sessions.

When deeper stats are available, there are interesting details like many more pages viewed from the (oddly few) tablets.

Apps are closer, and iOS more often (but not always) leads in App downloads, but analytical data is worse here generally so it’s hard to say how much /use/ we get per platform.

Replies to Steven Hoober
Kaleberg replied:

I’ll bet ESPN has some interesting stats. A non-computer saavy friend of mine recently bought an iPad just to watch sports. He has a very nice TV set, but my guess is that’s for watching specific events. An iPad is for following, sort of like tuning a transistor radio to the ball game in the old days. (Yes, he uses wifi. His wife set it up. She’s the techie.)

Tonio replied:

Akamai’s clients notably include ad networks which distribute ads on a lot of sites, including long tail sites, so I suspect the Akamai data is pretty good.

Replies to Tonio
studentrights replied:

I think that’s a false assumption.

Android app developers primarily depend on ads for revenue while iOS app developers depend on direct sales. The proliferation of ads in Android apps will skew the data in Android’s favor.

If you take in to account that Android is outselling iOS and developers depend more on ads to generate revenue, clearly more Android users use their phone as feature phones.

Randy Magruder said:

Another point to consider. I have friends with iPads. Oddly, I rarely see with them at work. “Where is your iPad? Didn’t you bring it in?” Answer is invariably: “No, I bring my laptop back and forth for work…the iPad is on the coffee table”. Guess where that is? Wi-Fi. I don’t think iPads get out as much as people think. What percentage of iPad users bought a cellular capable version? And of that, what percent actually USE the cell radio on the iPad? The iPad is a data glutton and I think those that buy it KNOW it, and stick to Wi-Fi (the popup probably helps).

In the Android space, the majority of devices are phones. Very few Android tablets are cellular capable. Most are Wi-Fi, but if the average Android tablet is smaller, it’s probably used more as an e-reader, gaming, app using device and less as an online browser, especially when it’s outside the home (some of us tether to our Android phones, but I’m guessing we’re in a minority). So, the profile I would bet on is that most iPads stay at home and are glued to Wi-Fi, whereas most Android devices are phones, with data plans, and yes an unobtrusive Wi-Fi notification. I’ll go a step further. I often turn OFF my wi-fi, because when I get within range of a hotspot that requires authentication, I get joined to it, but all my stuff can’t connect because I haven’t gone into a browser and authenticated. I end up turning the wifi off in annoyance because I don’t need my phone auto-switching me to a connection that requires a web page authentication, particularly in the car! Then I get home, and I forget I’ve turned off the Wi-Fi!

Dave said:

I think the iPad is driving this as others have pointed out. That said, another possible theory: there are clearly times when WiFi isn’t available and your only choice is cellular. At that point you use the phone, it’s your only option. When you do have WiFi available, you can choose to use your phone, or perhaps your WiFI only iPad or your laptop, etc. It may be the case that iOS users tend to use their iPhone or iPad even when other devices are available, yet Android users are far more inclined to jump to something else to browse with, given a choice.

Brent Royal-Gordon said:

A third theory: iPhone users are happy to use their phone’s web browser even when they’re at home, but Android users prefer using a computer over their phone unless they’re on the go and don’t have another data connection.

Alex the Ukrainian said:

Jason, I think 3rd reason (and big one at that) is that many Android users don’t need to use WiFi – they’re on 4G networks. For example, T-Mobile’s 4G (even though technically not fast enough to be called 4G) is plenty fast. Verizon’s LTE is super-fast. AT&Ts LTE – same. Sprint’s WiMax and new LTE – also quite fast. So iOS users, having no choice but a rather slow 3G connection, prefer WiFi whenever possible, whereas for many Android users it’s just not worth it – their pages load fast enough anyway.

Look at sales data – Verizon sold more than a few million LTE devices just last quarter. That’s a LOT 😀

Replies to Alex the Ukrainian

Jason Grigsby (Article Author ) replied:

I thought about the 4G portion of it, but this disparity has been going on for quite some time. I believe it has been around since before the iPad was released which is why I’m not convinced the iPad is the core reason. At minimum, the disparity has been around since before Android was dominated by 4G phones.

Secondly, let’s say it is because of 3G vs. 4G access, the data still seems to suggest that it would be in Google’s best interest to get more Android users on Wi-Fi. Google should want people to use the device for browsing more frequently and there is evidence that suggests some correlation between Wi-Fi access and browser usage.

Maybe it is a red herring. But if Google hasn’t looked into it, then it definitely should.

Jason Grigsby (Article Author ) said:

FWIW, I’ve asked a friend at Akamai to come in and clarify what the tablet vs. mobile phone break down looks like and if it exhibits the same disparity. So hopefully we’ll know more soon.

Alex the Ukrainian said:

PS: This comes from personal experience, too. I love my iPhone but gee is 3G slow. I therefore use WiFi whenever I can. When I play with friends’ Androids on LTE, they are just as fast as my iPhone is on WiFi.

Can’t wait for LTE iPhone this fall!

Replies to Alex the Ukrainian

Steven N replied:

FWIW: I get 7Mb/sec on my iPhone (AT&T) and see no difference in page load usage compared to my friends with LTE.

I think this alps boils down to another issue and that is battery life. Everyone of my friends using Android manage their radios like GPS, WiFi, 3G, LTE and such. Icons to turn them on and off so they can get their batteries to last the day. None of my iPhone friends do this.

I think many Android users simply don’t turn on the WiFi as much and iOS handles the transition between WiFi and cellular much more smoothly.

Replies to Steven N
Arah replied:

Chiming in on the battery issue, here. I went from an iPhone 3G to an Android phone in December. Unless I know I’m connected to WiFi that can keep my connection (i.e. at home), I turn the WiFi seeker on my Bionic off. Many places, including where I work, have iffy/wobbly WiFi, and my phone will use up all of its juice, trying to maintain the connection. So, Verizon 4G LTE wins.

Greg said:

Another (small) reason for the disparity could be people like me. I have an Android phone, and I will use it when I need to on a cellular network. But as soon as I’m in WiFi range, I switch to my iPod Touch. (My carrier didn’t have the iPhone until recently.)

chibimagic said:

Another theory: wifi is much less power efficient on Android than on iOS, so Android users tend to leave it turned off, even when they’re within range of wifi networks. Meaning Android users do most of their browsing over cellular, while iOS users switch back and forth depending on network availability.

Skip said:

To Joe’s point and countering Alex’s point about 4G vs 3G, there are supposedly nearly twice as many Android phones as iPhones. Given that they are basically equal in terms of cellular web usage (if the Akamai data is correct), it’s pretty clear that, on average, Android users use the Internet far less than iOS users.

Jason Grigsby (Article Author ) said:

FWIW, Q1 2010, before the iPad was released, was the first quarter that Android outsold the iPhone:
http://www.npd.com/press/releases/press_100510.html

According to Nielsen at the time, Android had 9% of the installed base of phones versus the iPhone’s 28%:
http://blog.nielsen.com/nielsenwire/online_mobile/iphone-vs-android/

Gartner showed a smaller gap between the platforms: 15.4% iOS vs. 9.6% Android.

Despite Android’s growth, the disparity between web browsing was already taking place:
http://marketshare.hitslink.com/operating-system-market-share.aspx?qprid=8&qpcustomd=1&qptimeframe=Q&qpsp=43&qpnp=1

The Q1 2010 numbers were 38% of mobile web browsing was done on iOS and only 3% on Android.

What does this tell us? The disparity in overall browsing cannot solely be blamed on the iPad. The disparity existed before the iPad was released in April 2010.

What doesn’t it tell us? Whether or not the difference in Wi-Fi versus cellular usage exist both in mobile phones and tablets. I hope we’ll have that data soon, but I don’t think it is as clear cut as people think.

Matthew Smith said:

In my opinion, this shows that Android owners use the phone when there’s no other option – when they’re away from wi-fi. In my experience, people with iPhones will use them at home even if their laptop is nearby. I even do it myself, as there are some apps on my iPhone which are quicker for getting certain info, such as TV listings, sports scores, and weather.

I imagine the ease of joining Wi-Fi networks on an iPhone or iPad is a big contributor too.

Rowan said:

Everyone in our family not old enough to own and pay for a mobile phone contract has an iPod touch, as their only computer. Lots of kids surfing YouTube, FaceTime etc.

John Harrison said:

What happened around July 25th that Mobile Explorer traffic doubled and iOS traffic dipped by about the same amount? Everybody decide to use their Nokias at once and then put them away?

latca said:

To me the higher wifi mobile safari traffic is due to the fact that most popular Android devices are phones, while the iOS ecosystem also includes many non-phone devices such as iPod Touches and iPads.

Jan said:

The reason for browser usage disparity over WiFi on iOS Devices has a name. It’s called Bonjour. Apples zero configuration networking just works ( most of the time ).

Neill said:

I think there may well be a disparity in how the analytics are read. I work in a web design agency, and I will often hear people of the analytics team say iPhone and iPad are predominantly accessing a site. What they fail to see is that due to all the variations of devices that run android, it can often be the most popular OS. I recently found this was the case for a large airline based in the UK.

I’m not saying that these stats have been misread in the same way, but with all the combinations of different browsers and devices there may be a large margin of error.

Replies to Neill

Steven Hoober replied:

While we hope these big organizations, with oversight from clever folks like Jason, are doing it well, I see a LOT of bad stats. Some is data gathering, but most is how you parse the info gathered.

Yours is a good one, but here’s another typical case I see. You gather device specifics using something (JS, say) that only gets stats for a few smartphones, and the major desktop browsers.

When the numbers are run, all the “we didn’t get the device” ones are gathered under an “other” category. This is, with the desktop-web mentality of most organizations assumed to be Opera or iCab or bots or something unimportant. And then, they build the visitor charts, especially for mobile, out of the KNOWN browsers.

You can see a pie chart for Mobile Use of iOS, Android, BB, and the numbers add up to 100%. That’s wrong. When I dig into it (and we grab other data), sometimes I find as much as 40% of visitors are on featurephones, or using Symbian, or something else crucial that was missed entirely.

Seriously, 40% featurephones with a site that didn’t really work on most of those browsers. How much money are they loosing by not targeting this market? Good stats matter.

Lith said:

I’m curious about one thing, though. Has anyone (a research body) conducted tests on Android and iOS to find out how much the device sends back and forth (either updates or app-based traffic or something users don’t even know about). I’ve often seen my router’s wifi light blinking profusely, and at that time my computer tends to be off and the only devices on standby are my girlfriend’s iPad and my iPhone4. I’m pretty sure no one in the neighborhood is on my network cos I live in the middle of nowhere (as in, I can look out the window and see nothing on all sides).

François Zöfel said:

I’m IT manager and we deployed a secure wireless in my organisation across 3 countries and we noticed that android 2.x smartphones were not working on WPA2 enterprise wifi networks.

The devices join the wlan, but for some reason, they never route their packets to the default gateway. My support team traced the problem to a faulty broadcom driver on the distribution used by the smartphones affected. Broadcom only resolved the bug on driver shipped with ICS and most smartphones could not be updated beyond 2.3, leaving them without wifi at work.

So I guess the difference you see is just that: people unable to use the wifi of their workplace because of their buggy android device, while cursing the IT department for the “incompatible” wifi.

Ted T. said:

Don’t most people turn off the “join this WiFi?” prompt on iOS? I know I do — it would be very annoying otherwise. I only turn it back on when traveling abroad in places where I have limited/no data roaming.

@Neill – iOS browsing has a 60% share — so no matter how Android devices get misidentified if you add up all devices that are not iOS, you only get 40% as the maximum Android share.

I think there is a much simpler explanation than a lot of the convoluted ones here — iOS devices are simply better. We are conditioned to seeing Android as the latest and greatest version running on a flagship devices, that offers a first rate user experience. But that isn’t what Android users are using — the vast majority of them are stuck on some Andoid 2.x or even 1.x device with doubtful battery life and an inferior web browser. Meanwhile a large majority of iOS devices are running the latest iOS and every new iPhone or iPad model sells more than all previous iPhone or iPad models combined. So most iOS users are running the latest software on the latest hardware and have an excellent experience.

Until at the very least Google starts updating Android OS directly rather than through carriers and hardware partners, who have a strong interest in blocking/delaying updates, Android will continue to be an also ran.

chapman said:

It may have been said before but, no wonder the phone companies push the Android phones over iphone. If it’s true that the Android users tend to use more cellular connections than wifi and connecting to a wifi hotspot is that much more difficult to connect to using Android, they in turn could use more data. That equals more money in the bank. Something to think about.

Guy Podjarny said:

Awesome post, Jason! I’m really happy to see Akamai IO data put to good use 🙂

There’s a lot of further breakdown to be done on Akamai IO’s data, and we’re working on sharing those further details too. In the meantime, here’s some insight into iPad/iPhone breakdown:

On non-cellular networks, Mobile Safari accounts for ~67% of mobile browser activity. iPad accounts for ~43% of that, and iPhone/iPod-touch for the remaining 24%. Android WebKit’s share is ~18%. So iPhone still had a notable lead over Android on non-cellular networks, but the gap isn’t as big if you don’t count the iPad.

On Cellular networks, Mobile Safari accounts for ~35% of mobile browsing, and only ~7% of that comes from iPads. Android WebKit’s share is ~38%, and less than 1% of that comes from tablets.

I’d suggest looking at these stats with a margin of error of 2% or so, especially for Android, due to the difficulty in accurately identifying the device.

Replies to Guy Podjarny

Jason Grigsby (Article Author ) replied:

Thanks Guy!

So if I’m reading what you wrote correctly, the breakdown for iPhone/iPod Touch versus Android phone looks like this:

On non-cellular networks
iPhone/iPod Touch: 24%
Android: 18%

On cellular networks
iPhone/iPod Touch: 28%
Android: 38%

So while top end numbers aren’t as dramatic (67% for mobile safari if you include iPad), the actual change is still there. In fact, Android phones go from being behind when it comes to Wi-Fi to ahead when it comes to cellular networks.

Another takeaway is the high percentage of traffic coming from iPads. Forty-three percent of Mobile Safari on Wi-Fi is nothing to sneeze at.

Don said:

Another thing that might be skewing those numbers is the iPhone and its ability to use AT&T’s wifi hotspots (at least at Starbucks) for free and without a clickthrough requirement. When I am near one my phone simply associates with it. So even if I would have been using the cellular data I end up using the wifi. Places where a popup demanding authentication happens I’m more likely to simply cancel and use the cellular data if I’m not going to be there long.

LG said:

I’m not sure if Wi-Fi and iPad explain all the story. Tablet web traffic is heavily dominated by iPad with around 90%, but iPad has a market share of around 68-70%.
So it seems there is still another reason why users on iOS devices consume a larger share of web traffic than their unit share.

Paul said:

First thing that came to mind when reading this – 3G on iPhones is SLOW. I use an Android on t-mobile and regularly get download speeds in excess of 8 megabits / sec – this makes mobile browsing a breeze. Then I see how long it takes my iPhoned friends to navigate to pages and I can’t imagine how they put up with it!

Charlie Clark said:

There are a couple of problems with Akamai’s data set: firstly, it only goes back to June 2012. This makes comparisons with other sources almost impossible until we have more data. The desktop stats for one do not come close to my own observations of a large corporate website.

Secondly, Akamai is dependent upon UA identification as opposed to the more sophisticated tricks of Adobe, Webtrends and Google. The problem with this is that Android browsers allow and even encourage users to mask the UA to stop some of the well-meant but dreadful mobile sites or app alerts (the BBC website is terrible for this). You can only avoid this using Javascript which Akamai doesn’t run. It wouldn’t surprise me if quite a few people have the UA set to desktop on their Android tablets.

Nevertheless, the salient points are: it’s great of Akamai to provide this information and let’s see how it develops; people “graze” more on tablets than they do on phones where they tend to “nibble” and the I-pad is still by far the dominant device.

Replies to Charlie Clark

Jason Grigsby (Article Author ) replied:

@Charlie, sorry to be the bearer of bad news, but Adobe, Google and likely WebTrends are all using UA detection.

For Adobe (meaning Omniture), I’ve been in a meeting with a senior product manager for the Omniture team talking to them about how they use DeviceAtlas. This fact is listed on the DeviceAtlas home page.

Google Analytics provides a mobile version of its analytics script that is meant to be run server-side to help with devices that don’t have JavaScript, so at minimum they are relying on UA detection there. For other Google properties, Google employees have told me that they use UA detection and maintain a large UA database.

StatCounter’s global stats use a UA database that they built up themselves. PPK has occasionally pointed out problems in their data, and I’ve encouraged them to go down the road and use fellow Dubliner’s DeviceAtlas system instead of rolling their own.

I can’t speak to WebTrends, but I sincerely doubt they are doing anything different.

As far as JavaScript is concerned helping distinguish UA spoofing, I would like to know how that would work and how reliable it would really be? In addition, what an incredibly complex moving target to keep track of. Now not only do you have to maintain a UA database, but you have to try to detect device characteristics that indicate the browser may be lying. All in JavaScript code that has to execute as quickly as possible and will be going on millions of web pages.

To summarize, I sincerely doubt any analytics provider is doing much sophisticated analysis to identify UA spoofing.

Charlie Clark said:

@Jason, you are, of course, right about UA ID strings, of which the various companies make a more or less good fist of identifying (Google and Webtrends currently fail spectacularly to give versions for Safari) but JS is still necessary to bust proxies. The point about desktop versus mobile UAs still stands, however. My tablet (Opera Mobile, Firefox) is configured to pretend to be a desktop most of the time unless I’m using a cellular connection when I either switch to Turbo or opt for the mobile view.

Comparability would mean comparing like with like – statcounter, etc. are not Omniture, Webtrends, Google, E-Tracker, ComScore, et al. Would be really nice to see Adobe and Google et al. releasing similar aggregated stats. At such a high-level there is little value in keeping them locked up.