Maitian: Why Douban is not a successful Web 2.0 site in China

Maitian dissects this little, most web 2.0-ish site in China and tells the truth.

In Oct, 2005, I was in the charge of product services at Tianya (a popular online community in China) when Douban went online. The other day, I talked to a VC about what could be the next nemesis of Tianya; he was a little awe-stucked when I told her the next one would be Douban, in stead of Mop.

I’ve never thought Mop was an opponent of Tianya, at the time when I was a Tianya guy. And now I’m saying what I said before: Douban is still not a big player, even though I’ve left Tianya. But I’m not alone. Just read Douban news stories, you will find the media has been irrationally sanguine about Mop fanfare, without caring about stats. The fact is: Mop is just as big as 2 sections at Tianya before being bought by Yizhou Chen (more is in my last May "Some stats about Mop" story), and there are a as many as 50 sections out there. Mop has got its big day since the acquisition.

Although Douban was launched about 2 years ago, and it didn't have as big traffic as Tianya did, as an ex-Tianya guy, I knew that Douban would be the biggest opponent out there. And I’ve talked much more about that in a January "Community 2007: Xie Wen, Keso, Maitian talk in the New Year" story.

I’m still optimistic about Douban, but  there are some lurking “website service structure” problems with it, and I had "the Power of Douban" and "Community: Scale and Services and Contents" stories for this. Here I’m trying to answer the question: Why is the Douban's  traffic curve  way too steep in 2005, but almost flat in 2006? For years of observations on Douban, personally I think the reason is the structural problem with Douban itself.

Douban's Alexa traffic ranking curve

The mere observation on traffic data didn’t render to me a full answer to the question, still I speculate that there are other factors working together to make a big difference on the curve  between in 2005 and 2006. I’m wondering what on earth those factors are.

It seems that all of this is going to be unlocked when I decided to get a study panel set up with He Tian, GinTonic, Si Jian (his advice about the panel), Hilaa. The mission of panel is that we release our case study report in which some exemplary websites will be examined in more than one aspect, thus collecting our useful data and experiences about running a site in Chinese website communities as a guide or reference book.

Luckily, Douban is our first case study. And I find myself surprisingly demystifying Douban Curve after giving Douban another analysis; the Curve mystery could lead you guys to the real truth about Doban.com

About traffic and service quality at Douban

The Alexa ranking curve of Douban, plus my experience about the traffic data at Douban.com, shows Douban can have a traffic as big as 200,000 with independent IP, 1,500,000 hits with PV. The traffic seems unimpressive, as some people might say. Hey, guys. You need to think again: Douban is just one-year old, with such an extent of traffic. It’s quite impressive for me. Of all the big traffic, many are exactly what Douban attracts with its high-quality contents.

A chart of Douban’s services is shown below:

Service type	Original Featured Service	community service	Social Networking Services	Spin-off services
Items	Book/music review/movie(reviewing and sharing)	groups	Neighbors	Nine-points blog

Among all services, OFS (Original Featured Service) is the vital part as the key services beside other services. How do the Douban users enjoy OFS? Some may answer like this:

I read a book about movie or music. Or I want to buy a book, but I’m not sure if I should do it. I can go to Douban for what others say about the book. 

I agree on the answer because I was a veteran book club member, and I know what those book buffs love to do. I assume that can be the conventional way people use Douban

Some people point that the OFS is superlative because, as they think, Douban guys have done much works on “algorithm” and “user behavior analyses”. I have the opposite views: Douban techies have “algorithm” and “user behavior analyses” innovated quite much, Nice job, guys. But, they were so focused on the two key factors that much little work was done on “user interaction”, which is a key build block of the whole service structure. Put it simply, “algorithm” matters more than “user interaction” to a site’s traffic.

Some of the readers might be shaking their heads. Well, you can post your comments below.  As an experienced observant, I want to point out that only OFS website architecture are unlikely to sustain as big traffic as 1,500,000 PV.

About Groups

It seems to me that Groups bring Douban.com the most part of traffic. Here’s why:

There are totally 25851 groups at Douban.com, of which group subscribers are listed below.

Subscribers	groups	total
More than 1000	240	240
500-1000	310	550
200－500	665	1215
100-200	735	1950
50-100	885	2835
20-50	1920	4755
10-20	1680	6435
5-10	3810	10245
3-5	4150	14395
2	4295	18690
1	7161	25851

Xba at MiracleNote has a short summary about the distribution stats on Douban Groups in the “Douban Groups are NOT the Longtail cases” piece. “We count the groups (with lower than 15 subscribers together), we got the number of more than 60, 000 subscribers, but we put the first groups (with more 500 users) together, the number is surprisingly 600,000, which is a steeper curve than statistical 28-dsitribution curve, the number of latter 80% groups is just 1/10 of first 2% (even not 20% ). I think the total number will be 90% of the whole group numbers” he writes.

Then, how  big  the  groups  with  big  portion  of  Douban’s  whole  traffic are? I did an analysis on the MoiveBuff (a red-hot group) stats yesterday, and found that there have been more than 6500 posts since it was created 700 days ago, equally, almost 10 posts in a single day. And I count the latest 200 posts and 200 greatest-hit pieces, it turns out there are so many responses among the cinephiles in the MovieBuff, the lasted 200 posts got 30 responses. The 200 greatest-hits have 20 comments.

It struck me as a new data continent that a most favorite group is so powerful that it can have as big traffic as a section in a large online community does. In the MoveBuff case, the post comments can be 1/5 of the total number of Movie Review at Tianya.

Remember that MovieBuff is just a small group at Douban

For many years, data analyses have made me know much more about the Douban traffic data than any others. So I did a math job, calculating the traffic of Douban Groups based on the Tianya traffic data, then compared with 150 PV of Douban. Here’s my extrapolation: groups brought a big part of traffic of 150 PV, I guess the exact number can be about 1, 000, 000. The whole traffic in the groups could add up to 70 % of the whole site.  And, I should note that 70% of traffic-- initial 240(with more 1000 subscribers) of more 25, 000 groups contribute-- sticks to the 28-distribution rule.

And one more thing to be noted: all Douban groups are much less about books/movies/music than about food and fashion and cosmetics. Basically, a Douban group discussion is about schmoozing about an interesting topic, in some ways the same as a post forum or a BBS.

About who  Douban users are

A Douban survey suggests that:

1. 80% Douban users are under 26, 90% are under 30. It’s a sure sign that Douban users are younger than average Internet users in China.

2. 50% users are students, 23% are company employees. Analysis: average Douban user is more well-educated than a Chinese Internet user.

Conclusion: Douban users are mainly young and well-educated.  

About the Douban Truth

Picture that if Baidu Post Forum has 70% of Baidu’s traffic, then Baidu looks like a real search engine? The answer is obviously NOT. Douban is also the case. The fact is: As I just said, Doban is less about books. It’s the user behaviors as a big driving force that convert Douban into a group-like social networking site.  Douban is essentially nothing less than a big post forum whose users just have good education background.

Now, I’ve got a full answer to Douban traffic curve mystery. Unfortunately, Abei has been blind about the mystery. Since 2005, Abei  had  been  working  hard  on  new  featured  services  to  attract  more new  users  beyond  college/university  students.  However,  the  new  services  barely  worked,  they  were  still  big  magnets  to  the  student  crowd  because  those  services  catered  well  to  the  special  high  tastes  of  this  demographics  for  books,  movies  and  music. We understand that because it also reminds us of a kind of consumeristic mindset—“I have no money, but I can pretend to be a petit bourgeois hanging out at a classy community online where no one knows if I have no money or not.”—when we were students. Douban was unsurprisingly, easily cosseted by the young-well-educated crowd.  And the big media coverage, plus students’ buzz, combined to  give  Douban  the  growth  curve  a  big  push.  

However, when the the young-well-educated landed on Douban, they just ended up talking about odds and ends, what they needed is just a small group without too many JAVA applets or AJAX stuff. They could give Abei’s sophisticated “algorithm” ideas a short shrift as long as the simple groups were created. And Abei himself  was  absorbed  in  optimizations  of  Douban’s  products,  still  ignorant  of  what  the  exactly  users  wanted  to  get.  The  predictable  result  is:  Abei  was  virtually  isolated  from  the  psyche  of  the  users,  both  users  and  Abei  got  to  a  point  where  they  split  over  what  new  attracting  services  Douban  could  offer,  even  though  they  were  bonded  to  Douban.  It’s  signal  that  the  fanfare  in  2005  was  gone  forever,  and  Douban  per  se  was  almost  on  the  verge  of  a  failed  innovation—most  of  users  didn’t  try  out  other  services,  other  than  Groups.  Without  the  next  pushing  force,  the  Douban  growth  curve  doubtlessly, predictably,  went  flat.

A quick flashback at the evolution of web in China shows us that Douban is NOT an isolated case; the young well-educated  saved unsuccessfully Douban as much as Muzimei did for Blogcn and BlogChina, the same way Super Girls' Voice (the Chinese counterpart of American Idol) helped Baidu post forum; like Xu Jinglei is the life-saver of Sina blog, much as Zhangyu did for Yuqoo(the Chinese equivalent of YouTube). The hidden gaming rule is like this: Behind every great site there is a woman. And there will be the one for Douban, it’s the Abei’s algorithm philoposhy could see; He’s trying to make that women appear with practicing the philosophy, NOT waiting.(the translator's note: Blogcn, BlogChina and Sina blog are 3 big blog service providers (BSP) in China.)

Finally, I can boil this lengthy piece down to the real truth about Douban: It’s clustered schmoozing-the only thing Chinese web users, regardless of education background, are addicted to doing as part of their life at any social networking site from Tianya to Baidu to Douban to much more.

Something you need to know...

About Maitian (a.k.a.麦田): An experienced Internet observer, Maitian has been keeping his eyes on online communities and social networking sites in China. He is one of the most influential Chinese tech bloggers. Along with others, he recently set up a research panel--1cm Community Study Panel(一厘米社区学习小组), a site dedicated to exploring the featured websites, and publishing some high-quality articles about the case studies--at MaaYee community, an online community about IT/web research.

He also blogs at TechWeb, a tech community for Chinese IT professionals. More is available at his blog : http://maitian.blog.techweb.com.cn/

About Douban (豆瓣)

Found by Yang Bo(杨勃,a.k.a.Abei) on March 26^th, 2005, Douban is a social networking site about book, movie and music reviews. It’s considered to be the best Web2.0 example site in China. As of Novmber, 2006, Douban has more than 330, 000 registered users, according to Chinese Wikipedia

Douban on the web:

The Chinese homepage: http://www.douban.com

The English homepage: http://www.douban.net