Why is mobile attribution in China so hard?
Product Content Strategist
Apr 26, 2017
China is a mobile-first society, but mobile attribution has remained a black box for international companies, in particular for app marketers. Unable to track campaigns reliably, there has been no way to effectively measure ad campaigns and reinvest in the retargeting of high-value users. Though Chinese companies are expected to spend more money on digital advertising than on TV campaigns in the next year, international businesses have had little success in stepping into one of the most promising economies in the digital world. A fragmented, messy market, issues with traffic congestion, low retention rates, mobile piracy and the Great Firewall all contribute to the current state of affairs. A more in-depth look at those hurdles reveals just how great an obstacle course stands in the way of most apps looking to head East.
The case for China
For mobile developers, the potential of the Chinese market has long been frustratingly out of reach. In a country where there are 600m smartphone users (a number expected to grow by 50m annually), and with a total population of 1.3bn (meaning market saturation is a long way off) only 20 percent of China’s top performing apps originate from outside of the country.
Despite this, international developers are eager to break into the Chinese market, and with good reason. China recently overtook the US to become the largest market in the world in App Store revenue, earning upwards of $1.7bn in Q3 of 2016. Chinese consumers spend more than Americans do on in-app purchases, even though they spend roughly the same amount of time in apps (on average, smartphone users in China spend five times more than they spent two years ago). Infrastructure is beginning to catch up as well - China recently began rolling out 4G/LTE to replace clunkier 3G/2G networks.
Traffic jam at the Great Firewall
Started in 1996, China’s web barrier, known colloquially as the Great Firewall (which keeps Facebook, Twitter, Pinterest, and many Google products from reaching Chinese users), has morphed from a list of blocked websites to a series of firewalls and proxy servers that filter out banned keywords, stop traffic to specific blacklisted pages, control social networking sites, and block access to VPNs. Officially known as the Golden Shield project, some of the tactics used to censor China’s internet include blocking access to specific IP addresses, poisoning DNS caches with false addresses and filtering URLs. These create a user experience that can be frustratingly opaque (for example, it can be impossible to tell whether you’re experiencing an actual network connectivity problem or are trying to access a banned site). Users may sometimes have access to a page or social media message that is blocked to others, in order to mask the degree of restriction.
In addition to censorship, the Great Firewall also slows mobile user experience with timeouts and dead ends. Because it controls everything that goes in and out of China on the web, it limits the bandwidth available for international apps going into China. This creates a lot of congestion, network delays and failure.
How the Great Firewall affects the mobile market
When it comes to mobile measurement, tracker URLs can take many seconds to redirect (and sometimes are not redirected at all) and packages from an app’s SDK are regularly dropped entirely, making it unreliable as well as slow. On a typical mobile network, most app latency happens during the last mile. In China, an app’s contents must travel far further as well as backwards and forwards through the firewall, which can ruin the experience of an app’s images, videos, and advertisements.
This has led to low retention rates. In the mobile ecosphere, retention is considered by many as the top metric for long-term success. China’s slow loading times and poor in-app performance mean that almost no apps (no matter the vertical) retain more than a quarter of their users after the first week.
The percentage of retained users over the first seven days is even lower for photo-editing apps (15.5 percent) and video apps (14.8 percent). Overall, a far greater percentage of Chinese smartphone (37 percent) users report opening an app only once (meaning a single app open) than users in the United States, where half of all users report opening an app at least 11 times.
Navigating a fractured ecosystem
In the US and EU, there are two main stores - the Apple Store and the Google Play store. In China, while iOS functions much the same, Google Play is banned. However, Android phones make up 76 percent of the market (in 2017, combined Android stores revenue are predicted to surpass worldwide iOS for the first time). In place of Google Play, there are numerous third-party app stores to take its place - by some estimates, there are over 400 different distribution platforms. One of the largest stores belongs to Tencent, China’s largest web portal. Phone manufacturer Huawei has its own app store, as do search engines Baidu and Qihoo.
Each app store requires its own infrastructure, so an app must create what’s known as an APK (Android package kit) in order to set up tracking for each one. This means that a developer has to create various app versions and also set up the APK with numerous custom events over and over again. Once they’ve been set up, it’s impossible to compare attribution data from different stores in a unified dashboard - with this setup, the data from each store is split over multiple apps and cannot be conveniently compared.
Privacy and piracy concerns
While there are local free solutions within China, some have historically approached attribution with a different vision of data property. Some local attribution providers profit from clients by selling data, leaving local app developers to decide between attribution analytics and data privacy. In addition, mobile piracy is rampant. In fact, it’s possible that if your app is a game, it’s already been pirated and is currently being played in China, without your permission and without any user paying for it.
Many of the stores also function like a network, so an app maker must pay to advertise (with stores taking 30 to 40 percent of revenue) and receive attribution analytics from the same source, rather than a third-party. For app marketers looking for a neutral attribution provider, options on the ground are thin.
Some attribution providers in China are perfectly legitimate. However, they require a server-to-server setup, which can be both cumbersome and miss key data points. With a server-to-server connection (rather than an SDK), the attribution provider uses a custom-built SDK to talk to the client’s server, through which the client can communicate with the attribution provider, although the client is never directly connected. This type of connection puts a great burden on the client and can miss key information that gives color and depth to attribution analytics, such as localized IPs, correct timestamps, and user agents.
In short, the mobile ecosystem in China promises a difficult but rewarding challenge. An attribution provider looking to enter the market would have to establish a way to be fast and reliable in an ecosystem that optimizes for neither.