China Web 2.0 blog recommends Qihoo for searching BBS (and I do too)

China Web 2.0 blog recommends using Qihoo as a tool to search conversations on BBS. We have been looking at Qihoo again the last month or so, and indeed, it is much improved since I wrote previous recommendations about BBS search engines. Zhongsou is also improved, though I think Qihoo seems to be more accurate and covers more posts.

One thing to keep in mind for companies who are seriously interested in tracking conversations is that while Qihoo, Teein, Daqi etc. are useful to get a feel for what is being said, they are not sufficient to obtain a deep understanding of how these conversations are impacting your business.

Here's how we approach work with our clients:

Find/collect messages:

This is the main function of search engines...helping you find what you are looking for.

1. Find enough conversations--search wide

For a study to be accurate and reliable, you need to make sure you have collected a significant, comprehensive or at least representative number of messages about the topic or topics you are studying. From our analysis, while the BBS search engines are improving, they still miss signifcant chunks of messages from big sites and especially from the "long tail" sites which may have less traffic, but which exert greater influence within certain communities which may be important to our cliets. For some clients, half our data can come from the top 20 BBS sites, while the other half of our data can be spread out over thousands of sites.

2. Find enough conversations--search deep

Certain industries such as automobile, consumer electronics, mobile phones and others have dedicated communities which can produce hundreds of thousands if not millions of messages a month. For example, one mobile phone site we monitor generates 500,000 messages. BBS search engines currently do not dig deep to get these conversations. And even if they did...what would you do with 500,000 results?

Make sense of the messages:

The next question is, once you have all this data, what can you do with it? While it is helpful to know how many conversations there are (buzz volume), don't you want to know what these conversations are saying? And if they are saying good things or bad things (sentiment)? Currently, this is beyond the scope of search engines (though some like Opinmind are trying in English. However, look closely and you will see the results are mixed).

If you are a company which generates few conversations, this is not a problem--read the messages to get a feel. But what if your company or industry has thousands, tens of thousands, or hundreds of thousands of messages every month?

This is where data mining/natural language processing tools come in. Companies like Nielsen Buzzmetrics, MotiveQuest, Cymfony and CIC data all specialize in not only finding how many messages there are, but what the messages are saying, what they mean, and most importantly, what clients should do.

Ultimately, what we do is more than just "new media monitoring" where you find/count mentions like you might do in traditional media monitoring. BBS/Blogs are filled with consumer conversations expressing opinions and experiences. Finding or counting them is not enough...understanding them should be the primary goal.