Dr. Google and Mr. Android
[Google announced changes to Android management last week. With around 500 million current users now, the mobile operating systems seems set on its course of reaching 1 billion users by later next year. Android still has many troubles though including fragmentation, rising security concerns, ambivalence in its relationship with hardware partners and ongoing developer monetization issues. Sheer numbers may wash away these problems, but many questions remain.]
Last week important news came out of Android. Andy Rubin who had founded the company and later ran it after the acquisition by Google announced that he was moving on to unspecified new projects within Google. The company had very little to say on the matter beyond that, and the void was filled with speculation. Google made several other changes to management and products last week. The company provided a lot more background on these and they quickly took up the headlines.
In the end, we were left with many questions about what exactly is going on at Google. I made several inquiries around the ecosystem but no one else seemed to know what was happening at Android. The new head of Android, Sundar Pichal, will continue to run the Chrome unit as well. This caused the blogosphere to speculate that Chrome and Android would someday be merged. This is possible, albeit unlikely any time soon. More to the point, it really seems that there is more to this story. We still do not quite understand what took place inside Google, but it seems that this marks a good time to review the Android operating system (OS) and its prospects.
Mr. Rubin is leaving Android in good shape. The company reported that the OS has had 800 million activations to date. Using my (so far accurate) model, this implies daily activations around 1.8 million. A healthy clip, albeit growth may be slowing a bit simply as a result of the numbers getting so big.
Source: Google and D/D estimates
This year will see some important changes in the Android marketplace. By my estimate roughly one third of Android devices that will ship this year will come out of the branded Chinese handset makers (the artists formerly known as the “shanzhai”). I put the estimates below together based on data from the chip makers, notably Mediatek, Qualcomm and Spreadtrum. I see this figure as a good proxy for “low-end” smartphones, those that retail for less than $150 at retail. I may be undercounting some of the low cost phones sold through prepaid channels in Europe and the US from the major global brands, but that is probably offset by the fraction of China branded phones shipped into those markets.
What should stand out from all this is that these so-called “low-end” phones are growing at much faster rate than the rest of Android. And my guess is that the 2013 estimates for these are probably still a bit too low.
The Android installed base is also growing very quickly. In their Rubin announcement Google noted that Android had already been activated on 750 million phones. The blogosphere commonly mistakes this figure for the installed base. In fact, that number is just the total devices sold to date, but plenty of those phones replaced existing Android devices. By the same model, the actual installed base is currently about 500 million devices, not a bad number. But bear in mind that the comparable figure for iOS is about 400 million, still very close.
Beyond the headlines
Despite these healthy numbers, there are five key vulnerabilities for Android – fragmentation, developer monetization, uneasy partners, security and business models.
We will not get into the fragmentation issue here. It’s been covered extensively in the past, and see no sign of it changing any time soon. Drop me a line if you want more on this.
A big question remains developer monetization. Just about every metric and survey out there shows a continuing discrepancy between the value of an iOS subscriber and that of an Android subscriber. Android users, on average, spend less on apps, view fewer ads, browse less traffic and spend less time on their device. This is not true for every app, but is true, by a wide margin, for the entire ecosystem.
The causes of this are a topic of hot debate within the industry. Theories include the weakness of the Google Play marketplace, the demographics of Android users, the quality of apps. All of the above play a role, and probably other factors as well. This is an important problem, and one that is still being explored. There are estimates of iOS monetizing anywhere from 2x to 10x better than Android per subscriber. Let’s assume that the number is 3x. That means there need to three times as many Android users than iOS users for the average app to be indifferent between the two platforms. By doing the math, Android does not reach that level until 2016.
Many factors could alter this equation. Android user growth in the next few years is going to be strongest in emerging markets, and monetization rates are much lower there. Tablet usage is growing and here the difference between monetization seems to even more heavily favor iOS. On the other hand, Apple could slow. My calculation above includes some assumptions about iOS growth continuing at a healthy clip. If Apple’s stock price is any indicator, this view is not widely held. Regardless, it will be many years before monetization on Android can approach that on iOS, and on a per user basis it seems unlikely that the gap will ever truly close.
The next concern is how hardware partners feel about Google. There are clear signs of growing unease here. Samsung’s Galaxy S4 launch last week never mentioned Google or Android. Most of the other Android phone makers are struggling to make any profit. Among the China-branded OEMs it would be tough to say there is a true love for Android. The larger vendors are doing what the carriers ask of them, and the smaller vendors build for whatever their chip suppliers can provide. Nor have the carriers ever been that fond of Android. They question Google’s long-term objectives and the company’s many side projects like Google Fiber, which are directly competitive. Add to this signs that the government in China could always make building Android phones less palatable for vendors there. The Chinese government and Google have a long-standing conflict, and this leaves Google little leeway to support handset makers there.
A Battle of Models
Nonetheless, the sheer numerical weight of close to 700 million Android phone shipments this year and over a billion next year could wash away most of these problems. Yet there is still one overriding concern that should weigh on the outlook for Android. At heart, Google and Android are not the same thing, and the way things stand now their business models are in many ways in direct competition.
Ask any Google employee in Mountain View and you will find that there is a distinct sense of Android being separate. They have their own buildings and their own cafeteria, which used to be closed to other Googlers. Under new management, this may change, but the sense of difference remains.
Google’s goal is to have as many people use its search as possible. One of the simplest, most elegant business models on the planet. Android’s model is much less clear. If you want to build an Android phone today, you can download the software, write all the hardware drivers, and you have a phone – but with one big exception. The Android team will not let you link that phone to the Google API – including Gmail, Google Maps and crucially the Google Play marketplace. In a bid to control fragmentation, team Android has used access to the Google API as a key inducement. And getting access to that API is not simple, requiring approval, a contract and various other hoops.
Google wants everyone to have access to that API, Android wants only Android devices to have that access. This explains why the Google apps for iOS are still so good, and why so many Google employees still carry iPhones. By my estimate, close to half of the â€˜mobile revenue’ that Google calls out every quarter comes from searches initiated on iOS devices.
This is not to say Android is a bad idea. When Google launched Android they realized they might face a world in which their search would be hindered on mobile. (Set aside for the moment that when they acquired Android their biggest fear was Windows Mobile.) And clearly their reliance on iOS could have been just as threatening. Launching Android, like Chrome on the browser, was a way to make sure they could maintain a voice in mobile and at the same time lower their traffic acquisition costs for mobile search. No one every really predicted that Android numbers would grow as quickly as they have.
So the company now faces a dilemma from which there is no easy solution. If they open up access to the API the effectively lose control of fragmentation. If they continue to restrict access to the API they limit avenues for other parts of Google’s business.
Of course, there are ways around this. You can buy a $45 Android tablet on the streets of Shenzhen, and it will have access to the full Google API. But if you try to trace back the source of that API key… let’s just say there are many things in China that are not fully clear. And this may be the third approach. Google may officially limit API access, but turn a blind eye to some of the cracks. An unfortunate side effect of this is that it creates some weird new flavors of fragmentation.
Which leads to what I think is going to emerge as a major headache for Android – the vulnerability of Android devices. Android devices have become magnets for spam and malware. There are ample studies both in industry (PDF) and academia (PDF) that point to this. There are many, many stories of apps that track users calls, download their data, and worse. From my conversations in China, it is clear that many phone manufacturers there are building software into their phones. These are generally benign or even helpful apps, a way for hardware makers to earn a few extra yuan. However, it demonstrates the ease with which a large number of handsets could be infected at a very deep layer.
At Mobile World Congress last month, I spoke to a number of smartphone security vendors. All of these offer solutions that are much safer than stock Android, but push most of them a bit and they will admit that if someone is really determined to act very badly and very technically, they can â€˜root’ Android devices beyond the ability of anyone to detect or curtail.
I do not want to come across as alarmist. There are vulnerabilities in home computers, and work laptops, in industrial systems and probably lots of other places we do not want to consider. The digital security industry has always been a race between the white and the black hats. Nonetheless, expect to hear a lot more about Android security in coming years.
Finally, I wanted to add a last-minute item to this list. Last week, Google performed another round of spring cleaning and killed off my favorite, most-heavily-used mobile app – Google Reader. You can read about why they did this on many blogs, but it is an important lesson. Free things cannot be relied on.
Would Google ever kill Android? This seems impossible today. One can safely predict with 100% certainty that this will not happen this year. But at some point, people have to ask this question. The handset makers have always questioned Google’s commitment. For most users, the answer is immaterial, they could just find another phone when their contracts are up or their phone dies. For the average app developer, the proverbial two guys in a garage, there is similarly little risk here. But if you are a major company and are investing heavily in mobile platforms, the longevity of a platform bears consideration.
Android is doing well. No question. And will continue to do well for many years. Again, no question. But that does not mean it is dominant, the way Windows was on the desktop for twenty or so years. There are plenty of vulnerabilities here, and we have not even gotten into the whole Amazon issue. But there is conflict with the relationship between Android and Google.
At heart, I think this conflict is a symptom of something bigger. Talk to most people in software today and they will preach to you the mantra of the web and HTML5. Up until very recently, most people in the industry assumed that HTML5 would someday remove the need for native operating systems. The idea being that web tools are widely used by millions of designers, and that eventually browsers will be good enough to power all apps. This is the underlying assumption behind Firefox OS. While this faith has wavered a bit lately, most people still believe it at some level. For Google, the company upon which so many fundamental web technologies emerge, we have to believe that this is the vision they are aiming for.
In a world in which the phone is just a browser, then Google wins all the search business it ever needs. In that vision, the need for a mobile OS eventually withers away. My sense is that many at Google still believe this is the ultimate outcome. (I used to be in this camp, but my faith, alas, has lapsed of late). And if that is where we are all headed, then Android and iOS and all the rest are merely transitional stages to be borne through, not eternal pillars of the firmament.
– Jonathan (@jaygoldberg)