Mei Guan Xi（没关系）

I realized that up until a couple months ago I still surfed English sites 80% of the time - which isn’t the smartest thing if I am trying to do an internet startup in China. Thus, in preparation for marketing our website, I forced myself to be a participant in the Chinese internet world, so here are some of my thoughts.

Before doing this startup, all the research told me that as far as the internet was concerned, information ruled in the US and entertainment in China. (New York Times, Keso-chinese, cnbeta-chinese, CWR) People point to things like demographics, 70% of internet users are under 30 in China versus 30% in the US (source: CNNIC) and lack of affordable entertainment as plausible causes for this behavior. Intuitively it makes sense, but to me, I was more interested in finding out if there was more detailed evidence to support that China is infact more entertainment based.

First order of business was to dig up some detailed data:

China Internet Network Information Center (CNNIC) is the official government organization that publishes all stats related to the internet in China. The latest report (in Chinese) they put out was in Jan 2008. FYI their methodology was based on telephone interviews (fixed and wireless users) and online surveys, at about 46k and 70k responses, respectively.

Here is what they say about what services Chinese people use on the internet (based on the last 6 months)

我发觉了如果想更了解中国互联网的情况我不应该老是上国外的网站，因此最近强迫了自己天天泡在各种中国的网站。 说实话确实有一点累，不过发觉中西方的网站差异还是不小。 创办了垦一垦之前我看过一些对中国互联网的统计及观点，比如(纽约时报-英文, Keso, cnbeta-chinese, CWR-英文). 这些消息的共识就是跟美国相对来说中国网民偏娱乐。据业士人分析，因素于网民年龄结构 — 中国网民70%小于30岁，美国只30% （来源：CNNIC）或没有别的划算的娱乐选择。 确实有一点道理，不过我还想看看有什么比较详细的数据可以证明中国网民只对娱乐性的网站感兴趣。

中国互联网信息中心 (CNNIC) 最新的互联网统计是1月发布的，调查方法为电话抽样调查（手机和固定电话加起来4.6万份）以及网上问卷调查（7万份）。以下是网络应用状况（6个月之内至少使用一次）：

As another data point, I found the following data on chinese usage patterns at Chinarank. They indicate that they have over 4 million users (toolbar users and partner data) to sample from and their methodology for this particular survey was based on random sampling of users of the top 10 websites of each of the 12 categories of sites. Report here-Chinese

我又发现了另外一家公司，中国网站排名，的网民日访问网站比例的调查。据该报告，中国网站排名拥有超过400万的用户数据（通过工具条及合作伙伴）可做抽样。 以下的图表来至于“《网站流量分析报告》中所筛选的12 类网站为对象，了解抽样网民群体在一段时间(以30 天为单位)内访问这些网站的频率，从而分析最经常被网民访问的网站类别。”

So far so good. I think this passes my gut test, but as I have learned, high-level summaries only tell part of the story.

As I dug deeper into the CNNIC data, it seems like the penetration of services are fairly even across China’s provinces – which seems a little surprising at first considering the disparity between the urban and rural parts of China. After giving it more thought, it actually does seem to make sense, while the absolute number of internet users might be night and day in urban vs rural, it seems as though if you’re an internet user in China, your surfing habits are quite uniform. Very interesting.

This is a graph showing search engine penetration across different provinces in China. Source: CNNIC (not sure why they made it a line graph vs a bar graph)

看了这些总结之后发觉跟我事先的猜测一致，不过有时候总结时常会忽视了一些有趣的发现。 详细研究了CNNIC的报告时我发现了所谓的网络应用使用率在各个区域都蛮相近的，因为往往听到东西部经济发展的差异当时有一点惊奇。然而想了一下，还是有一点道理；虽然东西部的网民数量的差异确实存在，但网民还是网民，不管你是中国哪里的网民，上网的习惯及爱好都差不了多少。

以下是在各个区域的搜索引擎使用率（来源：CNNIC – 不知道为何用了线图，而不用图标方块）

80%+ in Shanghai and 60%+ in Tibet. Pretty remarkable if you ask me.

没料到在西藏百分之六十的网名还用搜索引擎！

There is one exception to my previous statement – in online shopping!

网上购物明显的与其它的应用不一样。

在上海已达到了40%以上，山西的4倍! 美国网民在网上购物的穿透深度只有66%（来源：Pew Internet） 而且这个66%的比例不限制于过去六个月。 虽然这个差距比我想象的少多了，但是还要注意这个数据没提到平率（我信在这点美国网民对中国网民还是乘上几倍）

Look at how high it is in Shanghai! 40%+ while the lowest province is well less than a third of that at ~10%. Keep in mind that at 40%+, it’s not that far from the 66% in US (source: Pew Internet) , and even then the 66% from the US is not restricted to the past 6 months like the CNNIC study. So in reality the gap is even narrower. Not to be carried away with this data, this of course says nothing about the frequency of usage, which I guarantee is many times over in the US.

Aside from the CNNIC and Chinarank study, I wanted to try a different approach and see if I could glean any support from looking at web site traffic data.

Before you dismiss the next section as completely unreliable given well publicized statistical biases in different measuring services (particularly with services like Alexa), I believe in triangulating I arrived at a decent set of data. Not wanting to get into a long debate about why a site made the list or not, I am pretty confident that the trend illustrated below is a good reflection of reality.

另外，我还是想从网站流量的角度来思考这个问题。 首先，我了解个个网站排名都有自己的偏见及局限（尤其Alexa），但是通过交叉校验法，在以下的数据里应该体现得出来实际情况。

Yeah, that’s a lot of logos. Here was how I came up with the above picture.

I first pulled up lists from the following places for the US side:
Quantcast Top 50
ComScore Top 50 (Jan 2008)
Compete Top 20 (Dec 2007)
Hitwise Top 15 (Feb 2008)

They all have their own sampling methodologies whether it’s toolbars, ISP data, cookies, etc, but by cross-validating (ie. appearing on multiple lists) I felt like it was pretty likely that they really belonged on this list. About half of the US sites appeared on all 4 lists while the rest only appeared on 2 or 3, mostly due to the fact that beyond 20 or so, I only had the Quantcast and ComScore list to work with.

Ok, so getting the China data is trickier. As far as I know there isn’t any really reliable website ranking service in China, so my only alternative was to rely on Alexa. Alexa is notorious for being gamed, but I’ve also heard that they’ve cleaned up lately… so the first thing I did was to benchmark Alexa US rankings with the list I just came up with. Remarkably, of the top ones that appeared on all 4 lists above, all but 2 were on Alexa’s Top 50. If you look at the Alexa list, it’s clear that there is some gaming that’s still going on, but it doesn’t seem THAT bad. So armed with Alexa China’s semi-accurate Top 50 and cross-checking it with Chinarank’s Top 50 sites (which is probably biased too – but it was another data point), I arrived at the China list.

Of course you can argue that if a site gamed Alexa, it was probably trying to game Chinarank, too. Fair enough, but there was a decent enough disparity between the two that I found it to be a decent validator. If there is a better approach, I am open to trying it.

Before I start drawing some conclusions, I also want to make clear some of the shortcomings of this list. First is that portals offer a ton of services (blogs, mail, maps, news, videos, etc), thus, they clearly have an unfair advantage over more focused sites. Second, this list does not tabulate proportion of reach by category, which is to say that fewer sites in a category doesn’t necessarily equal less proportion of traffic for that category, but it’s a reasonable approximation.

首先，美国网站流量的来源以下：
Quantcast 50大
ComScore 50大 (1月 2008)
Compete 20大 (12月 2007)
Hitwise 15大 (2月 2008)

每个调查公司都有自己的抽样方法，不管是通过工具条，ISP，或 cookies，但是通过了交叉校验法我认为还是蛮可靠的。以上的一半的美国网站都出现在四个列表上；剩下的一半基本上是靠Quantcast和ComScore来判断的因为列表不全的问题.

如果要分析中美国的网站对比最好是用同一个调查来对比，这样才可以保证对比有意义。但很可惜这4家在中国没什么可用的数据，因此就要依赖Alexa了。虽然为了Alexa的排名一些网站会用不同的手段来作弊，我听说Alexa最近又改善了统计的算法。到目前为止，一看这个列表， 很明显一些偏见还是存在的。不过简单的验证一下，美国15大网站在Alexa 50大，只有两个不在。 因此我觉得在这个情况下，尤其加上了中国网站排名50大的对比（一定也有一些偏见），还算可以用。有可能不是百分之百准确，但是趋势因该是正确的。当然你也可以说如果一个网站在作弊，对两个排名的网站不是都会有影响吗？看了两个排名之后，发觉还是有一些差异，所以这个问题应该不会特别大。另外，我也不知道有什么比较好的方法。

在做我的总结之前还有一些东西要声明。首先，大多数的门户各种各样的应用都有，因此比一些垂直或专业网站占点优势。第二，在这个表里没有各个分类的流量比例，就是说在一个分类有了比较多的网站不一定代表占的流量就比较多，但是在这个情况下够接近了。

So with all the disclaimers out of the way, there are interesting conclusions we can draw from this data, some expected, some not:

• 1) It seems definitive that China is much more entertainment based when compared to the US (especially when you take into the account the differences between portals and blog styles in the US and China)
• 2) As expected e-commerce is much more mature in the US, but I was taken aback by the Shanghai data. I believe it’s a clear sign that e-commerce will only get more popular in China
• 3) It surprised me that sites like Verizon, AT&T, and Bank of America made the list. I associate e-commerce with shopping online, but I didn’t realize that so many service transactions are done online
• 4) China website format is portal dominate
• 5) Not surprising with the IM vs email usage in China. I’d hazard to guess US is the exact reverse

Things not easily gleaned from this data

• 1) News portals – even though a lot of people read news online like people do in the US, it’s more entertainment focused here, the politically charged stuff you see on some US news sites really don’t exist
• 2) China blogs are more diary-like and US blogs are more topical
• 3) I believe the piracy issue is a factor in China being entertainment oriented, just not sure how much. I’d guess video and music consumption in China is 90% pirated content (full movies/entire seasons of tv shows, baidu mp3, etc). Of course pirated stuff is rampant on youtube and torrents in the US but not nearly the scale it is here. As another data point, user-generated videos going viral is a fairly infrequent occurrence here (just my gut feel)
• 4) Competition from other media outlets is a big contrast. In the US, teenagers still watch A LOT of tv (source: Nielsen Nielsen, but it seems as though less and less young people here watch broadcast TV (predominately because the programs are generally regarded as of lower quality vs what they can get on the internet). Thus, I believe this leads people to view the internet as their main entertainment outlet
• 5) Haggling holding back e-commerce? I believe that’s a reason why things like price comparison sites in China have a hard time, but again, this is slowly changing as prices become more and more transparent
• 6) There is significantly less information put online in China. As a management consultant in the US, I remember looking up lexisnexis or factiva for a lot of data, but very tough to do here. Information exchange happens more at a personal level here. I believe this too is changing, but slowly

After looking at this data, it raises more questions. Any thoughts?

• 1) Why is the portal format so popular here in China?
• 2) I feel like diversity of information online is wider in the US, it’s just a gut feel, but not sure how to go able validating that or what possible causes could be?

总结：

• 1) 看了这几个图表之后，美国的确比中国偏信息（尤其考虑到中美国的门户内容及博客风格的差异）
• 2) 据我猜测，网上购物在美国果然比较成熟但看到了上海的比例还是令我大吃一惊。我觉得这个明显的表示网上购物肯定会越来越普及
• 3) 没料到Verizon, AT&T, and Bank of America会排得上名。 虽然我知道美国人经常在网上购物，没想到那么多人会在网上办手续
• 4) 美国网站的形式比较分散
• 5) 即时通信对电子邮箱 – 在美国我估计是倒过来的

有一些东西不容易在图里表示的：

• 1. 新闻门户 – 美国网民也一样的喜欢上网看新闻，不过内容还是有区别的。在一些美国网站很激烈的政治争论比较多
• 2. 博客在美国偏fixed topic，而在中国偏日记的形式
• 3. 在美国网上盗版的内容也很多（youtube, torrents）但还没到这边的程度。User Generated Content 相对来说还是多很多
• 4. 在美国电视的竞争力很强。 在美国很多青少年整天还是在看电视（来源：Nielsen ）而越来越多，在这边反而是颠倒。我发觉越来越少中国青少年在看电视，渐渐这个娱乐渠道由互联网来代替的
• 5. 美国人没有讨价还价的习惯 – 我认为这点是为什么在美国价格对比以及购物网站有办法做的起来的重大因素之一
• 6. 信息挂在互联网的习惯。 我记得在美国做咨询顾问的时候，要收集资料很简单，上一些网站比如lexisnexis, factiva, hoovers, 等等就很容易解决了。 这边我发觉大多数的信息还是通过个人关系所获到得。

最后做了这个分析，收获还是不少的，但是让我想到更多问题。 如果你有什么想法，请你跟大家分享一下

• 1. 为什么门户的形式那么普及？ 这点我有一点不太理解。不是说这边网站少，就是基本上都是为门户的形式
• 2. 个人感觉美国互联网的内容比较分散，各种各样的奇奇怪怪的网站都有，不知道怎么证明或因素会是什么